From mike503 at gmail.com Wed Jul 1 04:10:26 2009 From: mike503 at gmail.com (Michael Shadle) Date: Tue, 30 Jun 2009 17:10:26 -0700 Subject: What is the nginx equivalent of this? Message-ID: In Apache this is the example: SSLCipherSuite ALL:-ADH:+HIGH:+MEDIUM:-LOW:-SSLv2:-EXP Is this it? ssl_ciphers ALL:-ADH:+HIGH:+MEDIUM:-LOW:-SSLv2:-EXP; From delta.yeh at gmail.com Wed Jul 1 07:09:39 2009 From: delta.yeh at gmail.com (Delta Yeh) Date: Wed, 1 Jul 2009 11:09:39 +0800 Subject: What is the nginx equivalent of this? In-Reply-To: References: Message-ID: Does it looks alike? :) 2009/7/1 Michael Shadle > In Apache this is the example: > SSLCipherSuite ALL:-ADH:+HIGH:+MEDIUM:-LOW:-SSLv2:-EXP > > Is this it? > ssl_ciphers ALL:-ADH:+HIGH:+MEDIUM:-LOW:-SSLv2:-EXP; > > -------------- next part -------------- An HTML attachment was scrubbed... URL: From yazz.atlas at gmail.com Wed Jul 1 10:09:11 2009 From: yazz.atlas at gmail.com (Yazz Atlas) Date: Tue, 30 Jun 2009 23:09:11 -0700 Subject: Help with autoindex on; missing slash for directories Message-ID: <99e8cb160906302309q37c5b209o17fd68b1f52bbdb@mail.gmail.com> I'm working on upgrading some servers running nginx-0.6.34 to either nginx-0.8.4 or nginx-0.7.61. The problem is I have another tool that was crawling the nginx directory and following all the directories it found. With the latest version of nginx the the directories aren't given a slash so doing the automated crawl fails. The older nginx-0.6.34 output the following: ( Added pre tag to prevent auto formatting from happening.)



Index of /static/

Index of /static/
../
site/  30-Jun-2009 21:49                   -

The newer nginx-0.8.4 and nginx-0.7.61 out puts the following:



Index of /static/

Index of /static/
../
site  30-Jun-2009 21:49                   -

Notice the missing slash after the word site. I was relying on having the slash so I could navigate down directories with my crawler. Any one have a solution to add the slash back? Thanks! -Yazz Atlas From mdounin at mdounin.ru Wed Jul 1 13:30:47 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Wed, 1 Jul 2009 13:30:47 +0400 Subject: Help with autoindex on; missing slash for directories In-Reply-To: <99e8cb160906302309q37c5b209o17fd68b1f52bbdb@mail.gmail.com> References: <99e8cb160906302309q37c5b209o17fd68b1f52bbdb@mail.gmail.com> Message-ID: <20090701093047.GC5360@mdounin.ru> Hello! On Tue, Jun 30, 2009 at 11:09:11PM -0700, Yazz Atlas wrote: > I'm working on upgrading some servers running nginx-0.6.34 to either > nginx-0.8.4 or nginx-0.7.61. The problem is I have another tool that > was crawling the nginx directory and following all the directories it > found. With the latest version of nginx the the directories aren't > given a slash so doing the automated crawl fails. [...] > Notice the missing slash after the word site. I was relying on having > the slash so I could navigate down directories with my crawler. > > Any one have a solution to add the slash back? Which OS? Which filesystem? And could you please: 1. Test if this regression appeared in 0.7.40. 2. Show ./configure output on your system. Maxim Dounin From nginx-forum at nginx.us Wed Jul 1 14:23:12 2009 From: nginx-forum at nginx.us (miradev) Date: Wed, 1 Jul 2009 06:23:12 -0400 Subject: More subdomain mapping (dynamic 'root'?) Message-ID: I want to be able to map submains and urls to determined locations on disk, if they exist. The URL structure is something like: http://.admin.domain.com/ Where would letters, numbers and underscores , and is a domain name (e.g. http://cnn.admin.domain.com/www.cnnminiblog.com) Nginx should check if /var/www/clients//admin//html/ exists and use that as root: Some horrible pseudo code: server { listen localhost: 80; server_name *.admin.domain.com; if ($host ~ (*).admin.domain.com) { set $client = $1; if ($url ~ ^([^/]*)) { set $project = $1; set $myroot = /var/www/clients/$client/admin/$project/html; if ( -e $myroot ){ root $myroot; } } } ... } If this is not very 'nginxy', I am very open to alternate solutions. Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3586,3586#msg-3586 From info at pcintelligence.nl Wed Jul 1 15:59:31 2009 From: info at pcintelligence.nl (Michiel van Es) Date: Wed, 01 Jul 2009 13:59:31 +0200 Subject: question about nginx, slowloris and fastcgi Message-ID: <4A4B4FA3.4020205@pcintelligence.nl> Hi, i am new to Nginx and I am testing the latest stable Nginx as webserver with slowloris. When I test slowloris on my php enabled website I see Nginx running to 100% and get the following error: 500 Internal Server Error I don't think Nginx crashes (although the 100% cpu concerns me) but I think fastcgi/php is crashing. Are other people using Nginx and fastcgi/php are also experiencing this behaviour and is there a fix to ensure that my website is still running when slowloris is checking? Kind regards, Michiel From lists at ruby-forum.com Wed Jul 1 16:18:05 2009 From: lists at ruby-forum.com (Mahiti Support) Date: Wed, 1 Jul 2009 14:18:05 +0200 Subject: How to put rewrite rules for multiple plone instances In-Reply-To: <20090128103127.GG14282@rambler-co.ru> References: <441be1cb0901152328j7bef1a63t867334fc4335d60b@mail.gmail.com> <441be1cb0901160348w33a4c83es825c0b00819e80dd@mail.gmail.com> <20090116121345.GO39529@rambler-co.ru> <441be1cb0901160725k433982aar38b27196044f4b93@mail.gmail.com> <20090116154002.GS39529@rambler-co.ru> <441be1cb0901162232r5db93595veb891ef4b002c248@mail.gmail.com> <20090117080823.GW39529@rambler-co.ru> <441be1cb0901212051h158f5c0coaf5276cccf5a8ac4@mail.gmail.com> <20090122060941.GA19797@rambler-co.ru> <441be1cb0901280226s4b772356jd97991c631e0611a@mail.gmail.com> <20090128103127.GG14282@rambler-co.ru> Message-ID: Igor Sysoev wrote: > On Wed, Jan 28, 2009 at 03:56:53PM +0530, Vishnu Kumar D R wrote: > >> Hi Igor, >> >> Let me make it to the point >> >> whats the nginx equivalent for the statement in apache2? >> RewriteRule ^/(.*) >> http://localhost:8080/VirtualHostBase/http/%{HTTP_HOST}:80/$1 [L,P] > > location / { > proxy_pass > http://127.0.0.1:8080/VirtualHostBase/http/$host:80$request_uri; > } Hi Igor, Then what is it for RewriteRule ^/(.*) http://127.0.0.1:8050/VirtualHostBase/http/%{HTTP_HOST}:80/live/automiddleeast/VirtualHostRoot/$1 [L,P] Thanks Vishnu -- Posted via http://www.ruby-forum.com/. From mdounin at mdounin.ru Wed Jul 1 16:30:29 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Wed, 1 Jul 2009 16:30:29 +0400 Subject: question about nginx, slowloris and fastcgi In-Reply-To: <4A4B4FA3.4020205@pcintelligence.nl> References: <4A4B4FA3.4020205@pcintelligence.nl> Message-ID: <20090701123029.GE5360@mdounin.ru> Hello! On Wed, Jul 01, 2009 at 01:59:31PM +0200, Michiel van Es wrote: > Hi, > > i am new to Nginx and I am testing the latest stable Nginx as webserver > with slowloris. > When I test slowloris on my php enabled website I see Nginx running to > 100% and get the following error: > > 500 Internal Server Error > > I don't think Nginx crashes (although the 100% cpu concerns me) but I > think fastcgi/php is crashing. What's in error_log? Symptoms suggest that you've run out of file descriptors in your OS. At start nginx tries to warn you about the problem if worker_connections are more than open file resource limit, but even this isn't enough (since every connection usually allocates at leas 2 file descriptors, and system-wide limit for all processes isn't taken into account). Maxim Dounin > > Are other people using Nginx and fastcgi/php are also experiencing this > behaviour and is there a fix to ensure that my website is still running > when slowloris is checking? > > Kind regards, > > Michiel > > From artifexor at gmail.com Wed Jul 1 16:30:51 2009 From: artifexor at gmail.com (Artifex Maximus) Date: Wed, 1 Jul 2009 14:30:51 +0200 Subject: question about nginx, slowloris and fastcgi In-Reply-To: <4A4B4FA3.4020205@pcintelligence.nl> References: <4A4B4FA3.4020205@pcintelligence.nl> Message-ID: <9cbf3f070907010530p1e665fefy21cd0f6ea8696d@mail.gmail.com> Hi! On Wed, Jul 1, 2009 at 1:59 PM, Michiel van Es wrote: > Are other people using Nginx and fastcgi/php are also experiencing this > behaviour and is there a fix to ensure that my website is still running > when slowloris is checking? Check this thread: http://forum.nginx.org/read.php?2,3151,3151 Bye, a -------------- next part -------------- An HTML attachment was scrubbed... URL: From info at pcintelligence.nl Wed Jul 1 17:19:08 2009 From: info at pcintelligence.nl (Michiel van Es) Date: Wed, 01 Jul 2009 15:19:08 +0200 Subject: question about nginx, slowloris and fastcgi In-Reply-To: <20090701123029.GE5360@mdounin.ru> References: <4A4B4FA3.4020205@pcintelligence.nl> <20090701123029.GE5360@mdounin.ru> Message-ID: <4A4B624C.5050608@pcintelligence.nl> -------- Original Message -------- Subject: Re: question about nginx, slowloris and fastcgi From: Maxim Dounin To: nginx at sysoev.ru Date: 07/01/2009 02:30 PM > Hello! > > On Wed, Jul 01, 2009 at 01:59:31PM +0200, Michiel van Es wrote: > >> Hi, >> >> i am new to Nginx and I am testing the latest stable Nginx as webserver >> with slowloris. >> When I test slowloris on my php enabled website I see Nginx running to >> 100% and get the following error: >> >> 500 Internal Server Error >> >> I don't think Nginx crashes (although the 100% cpu concerns me) but I >> think fastcgi/php is crashing. > > What's in error_log? 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open files) 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open files) 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open files) 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open files) 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open files) 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open files) 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open files) 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open files) 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open files) > > Symptoms suggest that you've run out of file descriptors in your > OS. At start nginx tries to warn you about the problem if > worker_connections are more than open file resource limit, but > even this isn't enough (since every connection usually allocates at > leas 2 file descriptors, and system-wide limit for all processes > isn't taken into account). You're right but how can I stop this? Or how canI stop the slowloris attack..is it php fastcgi which is vulnerable to the slowloris DoS? > > Maxim Dounin Michiel > >> Are other people using Nginx and fastcgi/php are also experiencing this >> behaviour and is there a fix to ensure that my website is still running >> when slowloris is checking? >> >> Kind regards, >> >> Michiel >> >> > From mdounin at mdounin.ru Wed Jul 1 17:44:17 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Wed, 1 Jul 2009 17:44:17 +0400 Subject: question about nginx, slowloris and fastcgi In-Reply-To: <4A4B624C.5050608@pcintelligence.nl> References: <4A4B4FA3.4020205@pcintelligence.nl> <20090701123029.GE5360@mdounin.ru> <4A4B624C.5050608@pcintelligence.nl> Message-ID: <20090701134416.GF5360@mdounin.ru> Hello! On Wed, Jul 01, 2009 at 03:19:08PM +0200, Michiel van Es wrote: > > > -------- Original Message -------- > Subject: Re: question about nginx, slowloris and fastcgi > From: Maxim Dounin > To: nginx at sysoev.ru > Date: 07/01/2009 02:30 PM > > > Hello! > > > > On Wed, Jul 01, 2009 at 01:59:31PM +0200, Michiel van Es wrote: > > > >> Hi, > >> > >> i am new to Nginx and I am testing the latest stable Nginx as webserver > >> with slowloris. > >> When I test slowloris on my php enabled website I see Nginx running to > >> 100% and get the following error: > >> > >> 500 Internal Server Error > >> > >> I don't think Nginx crashes (although the 100% cpu concerns me) but I > >> think fastcgi/php is crashing. > > > > What's in error_log? > > 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open > files) > 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open > files) > 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open > files) > 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open > files) > 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open > files) > 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open > files) > 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open > files) > 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open > files) > 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open > files) > > > > > Symptoms suggest that you've run out of file descriptors in your > > OS. At start nginx tries to warn you about the problem if > > worker_connections are more than open file resource limit, but > > even this isn't enough (since every connection usually allocates at > > leas 2 file descriptors, and system-wide limit for all processes > > isn't taken into account). > > You're right but how can I stop this? See your OS tuning guides. Under FreeBSD use something like: sysctl kern.maxfiles=65535 sysctl kern.maxfilesperproc=60000 Note that other OS limits likely needs tuning too. > Or how canI stop the slowloris attack..is it php fastcgi which is > vulnerable to the slowloris DoS? When nginx with fastcgi_pass used in front of fastcgi - slowloris attack won't reach fastcgi. Maxim Dounin From nunomagalhaes at eu.ipp.pt Wed Jul 1 17:52:06 2009 From: nunomagalhaes at eu.ipp.pt (=?UTF-8?Q?Nuno_Magalh=C3=A3es?=) Date: Wed, 1 Jul 2009 14:52:06 +0100 Subject: 404 instead of "no inout file specified"? Message-ID: <6b1504c40907010652ib89ca65j40f918eb4035c3d5@mail.gmail.com> Greetings, If i provide a valid URL, it's processed. If i provide an invalid URL, nginx shows the 404 page except it the invalid URL ends in .php - in that case i get the "no input file specified" message. I'm in the process of changing my php files to an html extension (i assume that would "fix" things), however, i'd like to get a 404 instead of the message regardless. I tried using "fastcgi_intercept_errors off" in the conf for my server (in sites-available), but it didn't do anything (i did restart both nginx and the fast-cgi processes). Is there a better way? TIA, Nuno Magalh?es -- () ascii ribbon campaign - against html e-mail /\ ascii-rubanda kampajno - kontra? html-a retpo?to From roxis at list.ru Wed Jul 1 18:01:21 2009 From: roxis at list.ru (Roxis) Date: Wed, 1 Jul 2009 16:01:21 +0200 Subject: 404 instead of "no inout file specified"? In-Reply-To: <6b1504c40907010652ib89ca65j40f918eb4035c3d5@mail.gmail.com> References: <6b1504c40907010652ib89ca65j40f918eb4035c3d5@mail.gmail.com> Message-ID: <200907011601.21147.roxis@list.ru> On Wednesday 01 July 2009, Nuno Magalh?es wrote: > Greetings, > > If i provide a valid URL, it's processed. > If i provide an invalid URL, nginx shows the 404 page except it the > invalid URL ends in .php - in that case i get the "no input file > specified" message. > > I'm in the process of changing my php files to an html extension (i > assume that would "fix" things), however, i'd like to get a 404 > instead of the message regardless. > > I tried using "fastcgi_intercept_errors off" in the conf for my server > (in sites-available), but it didn't do anything (i did restart both > nginx and the fast-cgi processes). > > Is there a better way? > > TIA, > Nuno Magalh?es fastcgi_intercept_errors is off by default you need to set it on fastcgi_intercept_errors on; From mdounin at mdounin.ru Wed Jul 1 18:20:46 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Wed, 1 Jul 2009 18:20:46 +0400 Subject: 404 instead of "no inout file specified"? In-Reply-To: <6b1504c40907010652ib89ca65j40f918eb4035c3d5@mail.gmail.com> References: <6b1504c40907010652ib89ca65j40f918eb4035c3d5@mail.gmail.com> Message-ID: <20090701142046.GG5360@mdounin.ru> Hello! On Wed, Jul 01, 2009 at 02:52:06PM +0100, Nuno Magalh?es wrote: > Greetings, > > If i provide a valid URL, it's processed. > If i provide an invalid URL, nginx shows the 404 page except it the > invalid URL ends in .php - in that case i get the "no input file > specified" message. Looks like you've configured passing urls ending with ".php" to php for processing. So you got errors from php, not from nginx. It's expected behaviour, no? And, actually, "no input file specified" *is* 404 error. It's just has some fancy php-specific content. > I'm in the process of changing my php files to an html extension (i > assume that would "fix" things), however, i'd like to get a 404 > instead of the message regardless. > > I tried using "fastcgi_intercept_errors off" in the conf for my server > (in sites-available), but it didn't do anything (i did restart both > nginx and the fast-cgi processes). > > Is there a better way? If you want to change error page returned by php to something pretty-looking, you should: 1. Use "fastcgi_intercept_errors on;" to instruct nginx to intercept and change errors returned by fastcgi application. 2. Make sure you explicitly configured error_page for 404. Default compiled-in error pages are used only for nginx own errors, not for fastcgi_intercept_errors/proxy_intercept_errors. So your resulting config should look like: error_page 404 /404.html; location ~ \.php$ { fastcgi_pass ...; fastcgi_intercept_errors on; } ... Maxim Dounin From mdounin at mdounin.ru Wed Jul 1 18:28:37 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Wed, 1 Jul 2009 18:28:37 +0400 Subject: How to put rewrite rules for multiple plone instances In-Reply-To: References: <20090116121345.GO39529@rambler-co.ru> <441be1cb0901160725k433982aar38b27196044f4b93@mail.gmail.com> <20090116154002.GS39529@rambler-co.ru> <441be1cb0901162232r5db93595veb891ef4b002c248@mail.gmail.com> <20090117080823.GW39529@rambler-co.ru> <441be1cb0901212051h158f5c0coaf5276cccf5a8ac4@mail.gmail.com> <20090122060941.GA19797@rambler-co.ru> <441be1cb0901280226s4b772356jd97991c631e0611a@mail.gmail.com> <20090128103127.GG14282@rambler-co.ru> Message-ID: <20090701142837.GH5360@mdounin.ru> Hello! On Wed, Jul 01, 2009 at 02:18:05PM +0200, Mahiti Support wrote: > Igor Sysoev wrote: > > On Wed, Jan 28, 2009 at 03:56:53PM +0530, Vishnu Kumar D R wrote: > > > >> Hi Igor, > >> > >> Let me make it to the point > >> > >> whats the nginx equivalent for the statement in apache2? > >> RewriteRule ^/(.*) > >> http://localhost:8080/VirtualHostBase/http/%{HTTP_HOST}:80/$1 [L,P] > > > > location / { > > proxy_pass > > http://127.0.0.1:8080/VirtualHostBase/http/$host:80$request_uri; > > } > > Hi Igor, > > Then what is it for > RewriteRule ^/(.*) > http://127.0.0.1:8050/VirtualHostBase/http/%{HTTP_HOST}:80/live/automiddleeast/VirtualHostRoot/$1 > [L,P] Sorry, I'm not Igor, but probably something like location / { proxy_pass http://127.0.0.1:8050/VirtualHostBase/http/$host:80/live/automiddleeast/VirtualHostRoot$request_uri; } will work for you. Maxim Dounin From valery+nginxen at grid.net.ru Wed Jul 1 19:52:29 2009 From: valery+nginxen at grid.net.ru (Valery Kholodkov) Date: Wed, 1 Jul 2009 16:52:29 +0100 (BST) Subject: How to handle NGX_AGAIN returned by ngx_http_read_client_request_body() within handler module? In-Reply-To: <15d512b5e71760db17d3b689b07a7fee.NginxMailingList@forum.nginx.org> Message-ID: <6436495.72181246463549528.JavaMail.root@zone.mtgsy.net> Could you please show the code where you install the handler? ----- sayeo87 wrote: > Thanks guys for your replies. I have previously looked at how NGX_AGAIN is handled in the nginx source, and its along the lines of if ngx_http_read_client_request_body returns anything other than NGX_OK, return that from the module, otherwise return NGX_DONE. The problem is when I do this, my module is never called again. > > @Mike: I have tried what you described and it it promising in the sense that it appears that my handler function (2nd arg to ngx_http_read_client_request_body()) gets called again, but the handler function never returns to the main-line code of the module. Is this supposed to be expected behavior? > > I'm just confused because the way this is handled in the Nginx source does not seem to work, even though some people such as as http://markmail.org/message/knchl7ricknggvws have had success with it. -- Regards, Valery Kholodkov From yazz.atlas at gmail.com Wed Jul 1 20:48:29 2009 From: yazz.atlas at gmail.com (Yazz Atlas) Date: Wed, 01 Jul 2009 09:48:29 -0700 Subject: Help with autoindex on; missing slash for directories In-Reply-To: <20090701093047.GC5360@mdounin.ru> References: <99e8cb160906302309q37c5b209o17fd68b1f52bbdb@mail.gmail.com> <20090701093047.GC5360@mdounin.ru> Message-ID: <4A4B935D.3030803@gmail.com> On 7/1/09 2:30 AM, Maxim Dounin wrote: > Hello! > > On Tue, Jun 30, 2009 at 11:09:11PM -0700, Yazz Atlas wrote: > >> I'm working on upgrading some servers running nginx-0.6.34 to either >> nginx-0.8.4 or nginx-0.7.61. The problem is I have another tool that >> was crawling the nginx directory and following all the directories it >> found. With the latest version of nginx the the directories aren't >> given a slash so doing the automated crawl fails. > > [...] > >> Notice the missing slash after the word site. I was relying on having >> the slash so I could navigate down directories with my crawler. >> >> Any one have a solution to add the slash back? > > Which OS? Which filesystem? CentOS release 5.3 (Final) with ext3 filesystem, Linux example 2.6.18-128.1.10.el5 #1 SMP Thu May 7 10:35:59 EDT 2009 x86_64 x86_64 x86_64 GNU/Linux > And could you please: > > 1. Test if this regression appeared in 0.7.40. The problem still exists with 0.7.40. I did compiled 0.6.38 and it does not have the problem. However I would like to move to the either 0.7.x or 0.8.x. > 2. Show ./configure output on your system. [root at example nginx-0.7.40]# ./configure --http-proxy-temp-path=/tmp --prefix=/opt/nginx --with-http_ssl_module --with-http_gzip_static_module --http-fastcgi-temp-path=/opt/nginx/fastcgi_temp --http-proxy-temp-path=/opt/nginx/proxy_temp --http-client-body-temp-path=/opt/nginx/client_body_temp --group=nobody --user=nobody --prefix=/opt/nginx --with-http_realip_module --with-http_secure_link_module --with-http_stub_status_module --with-http_xslt_module/nginx/checking for OSp --group=nobody --user=nobody --prefix=/opt/nginx --with-http_realip_module + Linux 2.6.18-128.1.10.el5 x86_64 checking for C compiler ... found + using GNU C compiler0704 (Red Hat 4.1.2-44) + gcc version: 4.1.2 20080704 (Red Hat 4.1.2-44)--prefix=/opt/nginx --with-http_ssl_module checking for gcc -pipe switch ... foundastcgi-temp-path=/opt/nginx/fastcgi_temp --http-proxychecking for gcc variadic macros ... foundchecking for C99 variadic macros ... foundginx --with-http_realip_module --with-http_secure_checking for unistd.h ... foundatus_module --with-http_xslt_module checking for inttypes.h ... found checking for limits.h ... found checking for sys/filio.h ... not found checking for crypt.h ... found checking for Linux specific features checking for epoll ... found.40/ checking for sendfile() ... found checking for sendfile64() ... found checking for sys/prctl.h ... foundgure --http-proxy-temp-path=/tmp --prefix=/opt/nginx --wichecking for prctl(PR_SET_DUMPABLE) ... founddule --http-fastcgi-temp-path=/opt/nginx/fastcgchecking for sched_setaffinity() ... foundroxy_temp --http-client-body-temp-path=/opt/nginx/checking for poll() ... founddy --user=nobody --prefix=/opt/nginx --with-http_realip_module checking for /dev/poll ... not found checking for kqueue ... not found checking for crypt() ... not found checking for crypt() in libcrypt ... found checking for O_DIRECT ... found checking for F_NOCACHE ... not found checking for directio() ... not found checking for PCRE library ... found checking for OpenSSL library ... found checking for zlib library ... found checking for libxslt ... found checking for libexslt ... found checking for int size ... 4 bytes checking for long size ... 8 bytes checking for long long size ... 8 bytes checking for void * size ... 8 bytes checking for uint64_t ... found checking for sig_atomic_t ... found checking for sig_atomic_t size ... 4 bytes checking for socklen_t ... found checking for in_addr_t ... found checking for in_port_t ... found checking for rlim_t ... found checking for uintptr_t ... uintptr_t found checking for system endianess ... little endianess checking for size_t size ... 8 bytes checking for off_t size ... 8 bytes checking for time_t size ... 8 bytes checking for setproctitle() ... not found checking for pread() ... found checking for pwrite() ... found checking for strerror_r() ... found but is not working checking for gnu style strerror_r() ... found checking for localtime_r() ... found checking for posix_memalign() ... found checking for memalign() ... found checking for sched_yield() ... found checking for mmap(MAP_ANON|MAP_SHARED) ... found checking for mmap("/dev/zero", MAP_SHARED) ... found checking for System V shared memory ... found checking for struct msghdr.msg_control ... found checking for ioctl(FIONBIO) ... found checking for struct tm.tm_gmtoff ... found checking for struct dirent.d_namlen ... not found checking for struct dirent.d_type ... found Configuration summary + using system PCRE library + using system OpenSSL library + md5: using OpenSSL library + sha1 library is not used + using system zlib library nginx path prefix: "/opt/nginx" nginx binary file: "/opt/nginx/sbin/nginx" nginx configuration prefix: "/opt/nginx/conf" nginx configuration file: "/opt/nginx/conf/nginx.conf" nginx pid file: "/opt/nginx/logs/nginx.pid" nginx error log file: "/opt/nginx/logs/error.log" nginx http access log file: "/opt/nginx/logs/access.log" nginx http client request body temporary files: "/opt/nginx/client_body_temp" nginx http proxy temporary files: "/opt/nginx/proxy_temp" nginx http fastcgi temporary files: "/opt/nginx/fastcgi_temp" > Maxim Dounin > From admin at ampprod.com Wed Jul 1 21:44:39 2009 From: admin at ampprod.com (AMP Admin) Date: Wed, 1 Jul 2009 12:44:39 -0500 Subject: Advanced Maintenance In-Reply-To: <4A4B935D.3030803@gmail.com> References: <99e8cb160906302309q37c5b209o17fd68b1f52bbdb@mail.gmail.com> <20090701093047.GC5360@mdounin.ru> <4A4B935D.3030803@gmail.com> Message-ID: <013d01c9fa73$9c781bf0$d56853d0$@com> I see lots of maintenance page solutions but I would like to keep serving images and a few directories. Is that possible? Right now we have: if (-f $document_root/maintenance.php) { rewrite ^(.*)$ /maintenance.php last; break; } But we would like to make the privacypolicy.php page and the images & promo directories always available. That way people can always see our advertisers, images, and privacy policy. From jgarvin.lists at gmail.com Wed Jul 1 22:50:53 2009 From: jgarvin.lists at gmail.com (Jon Garvin) Date: Wed, 01 Jul 2009 12:50:53 -0600 Subject: ignoring proxy_read_timeout? Message-ID: <4A4BB00D.4030707@gmail.com> We're getting upstream timeout errors on certan requests that generate large reports. I tried setting proxy_read_timeout 300 in both the location and server sections of the nginx.conf file, but even after restarting nginx requests still return a 504 Gateway Time-out after exactly 60 seconds. Any ideas? -- http://www.5valleys.com/ http://www.workingwithrails.com/person/8078 From info at pcintelligence.nl Thu Jul 2 00:22:30 2009 From: info at pcintelligence.nl (Michiel van Es) Date: Wed, 01 Jul 2009 22:22:30 +0200 Subject: question about nginx, slowloris and fastcgi In-Reply-To: <20090701134416.GF5360@mdounin.ru> References: <4A4B4FA3.4020205@pcintelligence.nl> <20090701123029.GE5360@mdounin.ru> <4A4B624C.5050608@pcintelligence.nl> <20090701134416.GF5360@mdounin.ru> Message-ID: <4A4BC586.9050302@pcintelligence.nl> Maxim Dounin wrote: > Hello! > > On Wed, Jul 01, 2009 at 03:19:08PM +0200, Michiel van Es wrote: > >> >> -------- Original Message -------- >> Subject: Re: question about nginx, slowloris and fastcgi >> From: Maxim Dounin >> To: nginx at sysoev.ru >> Date: 07/01/2009 02:30 PM >> >>> Hello! >>> >>> On Wed, Jul 01, 2009 at 01:59:31PM +0200, Michiel van Es wrote: >>> >>>> Hi, >>>> >>>> i am new to Nginx and I am testing the latest stable Nginx as webserver >>>> with slowloris. >>>> When I test slowloris on my php enabled website I see Nginx running to >>>> 100% and get the following error: >>>> >>>> 500 Internal Server Error >>>> >>>> I don't think Nginx crashes (although the 100% cpu concerns me) but I >>>> think fastcgi/php is crashing. >>> What's in error_log? >> 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open >> files) >> 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open >> files) >> 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open >> files) >> 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open >> files) >> 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open >> files) >> 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open >> files) >> 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open >> files) >> 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open >> files) >> 2009/07/01 13:43:01 [alert] 25226#0: accept() failed (24: Too many open >> files) >> >>> Symptoms suggest that you've run out of file descriptors in your >>> OS. At start nginx tries to warn you about the problem if >>> worker_connections are more than open file resource limit, but >>> even this isn't enough (since every connection usually allocates at >>> leas 2 file descriptors, and system-wide limit for all processes >>> isn't taken into account). >> You're right but how can I stop this? > > See your OS tuning guides. Under FreeBSD use something like: > > sysctl kern.maxfiles=65535 > sysctl kern.maxfilesperproc=60000 > > Note that other OS limits likely needs tuning too. I have to tune CentOS file handles and network proc's right? And then it should protect me from slowloris? > >> Or how canI stop the slowloris attack..is it php fastcgi which is >> vulnerable to the slowloris DoS? > > When nginx with fastcgi_pass used in front of fastcgi - slowloris > attack won't reach fastcgi. What is the fastcgi_pass? Is it included in the nginx installation? I am using a rather old nginx install from the Epel repositories for CentOS. > > Maxim Dounin > Michiel From nginx-forum at nginx.us Thu Jul 2 00:24:38 2009 From: nginx-forum at nginx.us (vburshteyn) Date: Wed, 1 Jul 2009 16:24:38 -0400 Subject: memcached settings Message-ID: <28fc561b63ff103c4ab4a3fc9678f462.NginxMailingList@forum.nginx.org> Hi and thanks to all in advance, I am setting up my nginx to run with memcached. Now my issue is that when i set it up and turn it on i keep getting 404 error page. So i am thinking that i am using the wrong settings. How would i go about figuring out what the right value of the memcached key is? sorry but truwly a newb to linux/nginx scene. Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3617,3617#msg-3617 From jerome at loyet.net Thu Jul 2 00:50:46 2009 From: jerome at loyet.net (=?ISO-8859-1?B?Suly9G1lIExveWV0?=) Date: Wed, 1 Jul 2009 22:50:46 +0200 Subject: memcached settings In-Reply-To: <28fc561b63ff103c4ab4a3fc9678f462.NginxMailingList@forum.nginx.org> References: <28fc561b63ff103c4ab4a3fc9678f462.NginxMailingList@forum.nginx.org> Message-ID: <3bea96c40907011350s4b938fa1p405e78803882ae79@mail.gmail.com> it would be helpfull to give us your conf so that someone can help you. ++ Jerome 2009/7/1 vburshteyn : > Hi and thanks to all in advance, > > I am setting up my nginx to run with memcached. ?Now my issue is that when i set it up and turn it on i keep getting 404 error page. > > So i am thinking that i am using the wrong settings. > > How would i go about figuring out what the right value of the memcached key is? > > sorry but truwly a newb to linux/nginx scene. > > Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3617,3617#msg-3617 > > > From grails at jmsd.co.uk Thu Jul 2 01:01:38 2009 From: grails at jmsd.co.uk (John Moore) Date: Wed, 01 Jul 2009 22:01:38 +0100 Subject: Treat .htm files as PHP? Message-ID: <4A4BCEB2.8010203@jmsd.co.uk> I know how to do this with Apache, not so sure with nginx. One one of the websites I am hosting I want to use some PHP, but I don't want to rename the files from *.htm to *.php - I want the *.htm pages to be parsed as PHP (but for that site, or location, only). I should add that I have PHP set up and working fine with nginx, so that is not an issue. TIA, John From mike503 at gmail.com Thu Jul 2 01:18:03 2009 From: mike503 at gmail.com (Michael Shadle) Date: Wed, 1 Jul 2009 14:18:03 -0700 Subject: Treat .htm files as PHP? In-Reply-To: <4A4BCEB2.8010203@jmsd.co.uk> References: <4A4BCEB2.8010203@jmsd.co.uk> Message-ID: <0FD4C1CE-C908-4365-A5B2-BA3A2C14327F@gmail.com> Should be as simple as changing a php block and changing the extension to htm and passing it on to fastcgi like normal. I think. Sent from my iPhone On Jul 1, 2009, at 2:01 PM, John Moore wrote: > I know how to do this with Apache, not so sure with nginx. One one of > the websites I am hosting I want to use some PHP, but I don't want to > rename the files from *.htm to *.php - I want the *.htm pages to be > parsed as PHP (but for that site, or location, only). I should add > that > I have PHP set up and working fine with nginx, so that is not an > issue. > > TIA, > > John > > From mdounin at mdounin.ru Thu Jul 2 02:53:52 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Thu, 2 Jul 2009 02:53:52 +0400 Subject: Help with autoindex on; missing slash for directories In-Reply-To: <4A4B935D.3030803@gmail.com> References: <99e8cb160906302309q37c5b209o17fd68b1f52bbdb@mail.gmail.com> <20090701093047.GC5360@mdounin.ru> <4A4B935D.3030803@gmail.com> Message-ID: <20090701225352.GJ5360@mdounin.ru> Hello! On Wed, Jul 01, 2009 at 09:48:29AM -0700, Yazz Atlas wrote: > On 7/1/09 2:30 AM, Maxim Dounin wrote: >> Hello! >> >> On Tue, Jun 30, 2009 at 11:09:11PM -0700, Yazz Atlas wrote: >> >>> I'm working on upgrading some servers running nginx-0.6.34 to either >>> nginx-0.8.4 or nginx-0.7.61. The problem is I have another tool that >>> was crawling the nginx directory and following all the directories it >>> found. With the latest version of nginx the the directories aren't >>> given a slash so doing the automated crawl fails. >> >> [...] >> >>> Notice the missing slash after the word site. I was relying on having >>> the slash so I could navigate down directories with my crawler. >>> >>> Any one have a solution to add the slash back? >> >> Which OS? Which filesystem? > > CentOS release 5.3 (Final) with ext3 filesystem, Linux example > 2.6.18-128.1.10.el5 #1 SMP Thu May 7 10:35:59 EDT 2009 x86_64 x86_64 > x86_64 GNU/Linux > >> And could you please: >> >> 1. Test if this regression appeared in 0.7.40. > > The problem still exists with 0.7.40. I did compiled 0.6.38 and it does > not have the problem. However I would like to move to the either 0.7.x > or 0.8.x. I mean *appeared* in 0.7.40, i.e. problem exists in 0.7.40 but not in 0.7.39. But after some thinking it's most likely 0.7.15 when dirent.d_type checking was initially introduced (i.e. problem exists in 0.7.15 but not in 0.7.14). >> 2. Show ./configure output on your system. > > [root at example nginx-0.7.40]# ./configure --http-proxy-temp-path=/tmp > --prefix=/opt/nginx --with-http_ssl_module [...] > checking for struct dirent.d_namlen ... not found > checking for struct dirent.d_type ... found [...] You system has dirent.d_type, but it looks like it's something wrong with it. I've tried to reproduce problem on vmware image of CentOS 5.3 i386 - but failed, everythings works ok. I've even tried initial release version and fully updated one (with 2.6.18-128.1.16.el5 kernel), no problems found. It's either x64 only issue, or something wrong with headers on your server(s). Unfortenately I have no x64 version on hand to test it there, but people report that everything is ok at least on other x64 distributions. Could you please check your headers, most notably /usr/include/dirent.h and /usr/include/bits/dirent.h? Maxim Dounin From mdounin at mdounin.ru Thu Jul 2 02:59:57 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Thu, 2 Jul 2009 02:59:57 +0400 Subject: question about nginx, slowloris and fastcgi In-Reply-To: <4A4BC586.9050302@pcintelligence.nl> References: <4A4B4FA3.4020205@pcintelligence.nl> <20090701123029.GE5360@mdounin.ru> <4A4B624C.5050608@pcintelligence.nl> <20090701134416.GF5360@mdounin.ru> <4A4BC586.9050302@pcintelligence.nl> Message-ID: <20090701225957.GK5360@mdounin.ru> Hello! On Wed, Jul 01, 2009 at 10:22:30PM +0200, Michiel van Es wrote: [...] > >> Or how canI stop the slowloris attack..is it php fastcgi which is > >> vulnerable to the slowloris DoS? > > > > When nginx with fastcgi_pass used in front of fastcgi - slowloris > > attack won't reach fastcgi. > > What is the fastcgi_pass? Is it included in the nginx installation? > I am using a rather old nginx install from the Epel repositories for CentOS. Uhm, fastcgi_pass is nginx configuration directive to pass request processing to fastcgi backends. Maxim Dounin From mdounin at mdounin.ru Thu Jul 2 03:19:21 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Thu, 2 Jul 2009 03:19:21 +0400 Subject: Advanced Maintenance In-Reply-To: <013d01c9fa73$9c781bf0$d56853d0$@com> References: <99e8cb160906302309q37c5b209o17fd68b1f52bbdb@mail.gmail.com> <20090701093047.GC5360@mdounin.ru> <4A4B935D.3030803@gmail.com> <013d01c9fa73$9c781bf0$d56853d0$@com> Message-ID: <20090701231921.GL5360@mdounin.ru> Hello! On Wed, Jul 01, 2009 at 12:44:39PM -0500, AMP Admin wrote: > I see lots of maintenance page solutions but I would like to keep serving > images and a few directories. Is that possible? > > Right now we have: > if (-f $document_root/maintenance.php) { > rewrite ^(.*)$ /maintenance.php last; > break; > } > > But we would like to make the privacypolicy.php page and the images & promo > directories always available. That way people can always see our > advertisers, images, and privacy policy. Feel free to configure maintenance checks only in locations you need them. E.g. location / { # maintenance checks here ... } location /always-visible { # no maintenance checks here } Maxim Dounin p.s. Please start a new thread when asking unrelated questions. Just hitting "reply" is a bad idea. From avleen at gmail.com Thu Jul 2 03:22:22 2009 From: avleen at gmail.com (Avleen Vig) Date: Thu, 2 Jul 2009 00:22:22 +0100 Subject: Curious about old bug Message-ID: <33c66c80907011622ye551a2cna4abd7e936a06939@mail.gmail.com> I'm reading through the changelog from 0.6.35 -> 0.7.59, and saw this: Bugfix: now accept threshold depends on worker_connections I was wondering if anyone could tell me what that was? Thanks :) From nginx-forum at nginx.us Thu Jul 2 03:28:54 2009 From: nginx-forum at nginx.us (vburshteyn) Date: Wed, 1 Jul 2009 19:28:54 -0400 Subject: memcached settings In-Reply-To: <3bea96c40907011350s4b938fa1p405e78803882ae79@forum.nginx.org> References: <28fc561b63ff103c4ab4a3fc9678f462.NginxMailingList@forum.nginx.org> <3bea96c40907011350s4b938fa1p405e78803882ae79@forum.nginx.org> Message-ID: Basically we are running a nginx/apache2/django setup memcached works, i tested it and such. From here it goes to apache server. Everything works right up till i enable the set $memcached_key $uri; memcached_pass 127.0.0.1:11212; server { listen 10.176.67.151:80; server_name www.mysite.com; location / { proxy_pass http://10.176.67.151:8080/; include /etc/nginx/proxy.conf; set $memcached_key $uri; memcached_pass 127.0.0.1:11212; } location /static/ { alias /home/django/broadway.com/broadway/static/; expires 1d; } location /media/ { alias /home/django/broadway.com/src/django/django/contrib/admin/media/; expires 1d; } } Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3617,3629#msg-3629 From grangier at gmail.com Thu Jul 2 03:38:50 2009 From: grangier at gmail.com (xgdlm) Date: Thu, 2 Jul 2009 01:38:50 +0200 Subject: memcached settings In-Reply-To: References: <28fc561b63ff103c4ab4a3fc9678f462.NginxMailingList@forum.nginx.org> <3bea96c40907011350s4b938fa1p405e78803882ae79@forum.nginx.org> Message-ID: <0C817A63-E309-4717-985F-B174A96B6431@gmail.com> Hello, http://wiki.nginx.org/NginxHttpMemcachedModule server { listen 10.176.67.151:80; server_name www.mysite.com; location @fallback { proxy_pass http://10.176.67.151:8080/; include /etc/nginx/proxy.conf; } location / { set $memcached_key $uri; memcached_pass 127.0.0.1:11212; error_page 404 = @fallback; error_page 502 = @fallback; } location /static/ { alias /home/django/broadway.com/broadway/static/; expires 1d; } location /media/ { alias /home/django/broadway.com/src/django/django/contrib/ admin/media/; expires 1d; } } xav Le 2 juil. 09 ? 01:28, vburshteyn a ?crit : > Basically we are running a nginx/apache2/django setup > > memcached works, i tested it and such. From here it goes to apache > server. Everything works right up till i enable the > > set $memcached_key $uri; > memcached_pass 127.0.0.1:11212; > > > server { > listen 10.176.67.151:80; > server_name www.mysite.com; > > location / { > proxy_pass http://10.176.67.151:8080/; > include /etc/nginx/proxy.conf; > > set $memcached_key $uri; > memcached_pass 127.0.0.1:11212; > > > } > location /static/ { > alias /home/django/broadway.com/broadway/static/; > expires 1d; > } > location /media/ { > alias /home/django/broadway.com/src/django/django/contrib/ > admin/media/; > expires 1d; > } > } > > Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3617,3629#msg-3629 > > From xufengnju at sina.com Thu Jul 2 05:15:30 2009 From: xufengnju at sina.com (=?utf-8?B?WFVGRU5H?=) Date: Thu, 2 Jul 2009 09:15:30 +0800 Subject: =?utf-8?B?UmU6IG1lbWNhY2hlZCBzZXR0aW5ncw==?= References: <28fc561b63ff103c4ab4a3fc9678f462.NginxMailingList@forum.nginx.org> Message-ID: <200907020915247657603@sina.com> Hi, Make sure what you stored in memcached is exactly as what Nginx will retrieve from it. Yours, ------------------ XUFENG 2009-07-02 ------------------------------------------------------------- ????vburshteyn ?????2009-07-02 04:24:38 ????nginx ??? ???memcached settings Hi and thanks to all in advance, I am setting up my nginx to run with memcached. Now my issue is that when i set it up and turn it on i keep getting 404 error page. So i am thinking that i am using the wrong settings. How would i go about figuring out what the right value of the memcached key is? sorry but truwly a newb to linux/nginx scene. Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3617,3617#msg-3617 From avleen at gmail.com Thu Jul 2 05:23:06 2009 From: avleen at gmail.com (Avleen Vig) Date: Thu, 2 Jul 2009 02:23:06 +0100 Subject: Tuning workers and connections Message-ID: <33c66c80907011823r7c6224e8w2945765c150f1f55@mail.gmail.com> Hi folks, I have some questions about tuning nginx for best performance, on a site which handle ~1000 requests per second. In our config, we have: worker_processes 16; worker_rlimit_nofile 32768; events { worker_connections 8192; } This is on a server with 8 CPU cores and 8Gb RAM. My understanding is that this should allow nginx to establish up to 32k network connections, because that is the limit of worker_rlimit_nofile. We serve very few files off the disk, maybe a dozen CSS and JS files. Everything else is handed to backends using proxy_pass. So with one fd for the browser, and one fd to the upstream, we should be able to handle 16k concurrent connections. However what we're seeing, is that around 5k connections we get a performance hit. Connections slow down, take longer to establish, etc. The load on the box is almost zero, nginx is clearly working very fast and efficiently, but I'm not sure why it slows down. Any thoughts? Also, I'm thinking about enabling the multi_accept option, but I couldn't find much documentation on how this works. It sounds like a high-performance tweak which we should be using if we get this many requests per second. Could it be that not using multi_accept is the problem? Nginx otherwise has to handle the incoming connection requests in serial and this is a bottleneck? Thanks :) From gabe at vtunnel.com Thu Jul 2 06:39:25 2009 From: gabe at vtunnel.com (Gabriel Ramuglia) Date: Wed, 1 Jul 2009 19:39:25 -0700 Subject: Tuning workers and connections In-Reply-To: <33c66c80907011823r7c6224e8w2945765c150f1f55@mail.gmail.com> References: <33c66c80907011823r7c6224e8w2945765c150f1f55@mail.gmail.com> Message-ID: <932ea6c90907011939k3eec710dla9c0d839f2cafc12@mail.gmail.com> I'm no expert, but if your load level is low, it does sound possible that you are having issues accepting requests in serial for each worker connection. I would try increasing the number of worker processes to 32 and see if that helps. Also, if you're seeing 5k connections, you should try increasing the number of worker connections to double what it is now, just in case. Be happy to hear any dissenting opinions. -Gabe On Wed, Jul 1, 2009 at 6:23 PM, Avleen Vig wrote: > Hi folks, I have some questions about tuning nginx for best > performance, on a site which handle ~1000 requests per second. > > In our config, we have: > > ? ?worker_processes ?16; > ? ?worker_rlimit_nofile 32768; > ? ?events { > ? ? ? ?worker_connections ?8192; > ? ?} > > This is on a server with 8 CPU cores and 8Gb RAM. > My understanding is that this should allow nginx to establish up to > 32k network connections, because that is the limit of > worker_rlimit_nofile. We serve very few files off the disk, maybe a > dozen CSS and JS files. Everything else is handed to backends using > proxy_pass. > So with one fd for the browser, and one fd to the upstream, we should > be able to handle 16k concurrent connections. > > However what we're seeing, is that around 5k connections we get a > performance hit. Connections slow down, take longer to establish, etc. > The load on the box is almost zero, nginx is clearly working very fast > and efficiently, but I'm not sure why it slows down. > Any thoughts? > > > Also, I'm thinking about enabling the multi_accept option, but I > couldn't find much documentation on how this works. > It sounds like a high-performance tweak which we should be using if we > get this many requests per second. > Could it be that not using multi_accept is the problem? > Nginx otherwise has to handle the incoming connection requests in > serial and this is a bottleneck? > > > Thanks :) > > From admin at ampprod.com Thu Jul 2 07:06:11 2009 From: admin at ampprod.com (AMP Admin) Date: Wed, 1 Jul 2009 22:06:11 -0500 Subject: Advanced Maintenance In-Reply-To: <20090701231921.GL5360@mdounin.ru> References: <99e8cb160906302309q37c5b209o17fd68b1f52bbdb@mail.gmail.com> <20090701093047.GC5360@mdounin.ru> <4A4B935D.3030803@gmail.com> <013d01c9fa73$9c781bf0$d56853d0$@com> <20090701231921.GL5360@mdounin.ru> Message-ID: <024b01c9fac2$0e91b7c0$2bb52740$@com> I don't understand. How do we make the "no maintenance checks"? sorry, I'm still learning. If we want everything but privacypolicy.php, images directory and promo directory to go to maintenance how would that look? It seems that this puts the entire site in maintenance mode: Location / { if (-f $document_root/maintenance.php) { rewrite ^(.*)$ /maintenance.php last; break; } } I'm not sure how to do the no maintenance part. -----Original Message----- From: owner-nginx at sysoev.ru [mailto:owner-nginx at sysoev.ru] On Behalf Of Maxim Dounin Sent: Wednesday, July 01, 2009 6:19 PM To: nginx at sysoev.ru Subject: Re: Advanced Maintenance Hello! On Wed, Jul 01, 2009 at 12:44:39PM -0500, AMP Admin wrote: > I see lots of maintenance page solutions but I would like to keep serving > images and a few directories. Is that possible? > > Right now we have: > if (-f $document_root/maintenance.php) { > rewrite ^(.*)$ /maintenance.php last; > break; > } > > But we would like to make the privacypolicy.php page and the images & promo > directories always available. That way people can always see our > advertisers, images, and privacy policy. Feel free to configure maintenance checks only in locations you need them. E.g. location / { # maintenance checks here ... } location /always-visible { # no maintenance checks here } Maxim Dounin p.s. Please start a new thread when asking unrelated questions. Just hitting "reply" is a bad idea. From edhoprima at gmail.com Thu Jul 2 07:37:47 2009 From: edhoprima at gmail.com (Edho P Arief) Date: Thu, 2 Jul 2009 10:37:47 +0700 Subject: Advanced Maintenance In-Reply-To: <024b01c9fac2$0e91b7c0$2bb52740$@com> References: <99e8cb160906302309q37c5b209o17fd68b1f52bbdb@mail.gmail.com> <20090701093047.GC5360@mdounin.ru> <4A4B935D.3030803@gmail.com> <013d01c9fa73$9c781bf0$d56853d0$@com> <20090701231921.GL5360@mdounin.ru> <024b01c9fac2$0e91b7c0$2bb52740$@com> Message-ID: On Thu, Jul 2, 2009 at 10:06 AM, AMP Admin wrote: > I don't understand. ?How do we make the "no maintenance checks"? ?sorry, I'm > still learning. > > If we want everything but privacypolicy.php, images directory and promo > directory to go to maintenance how would that look? > > It seems that this puts the entire site in maintenance mode: > Location / { > ? ? ? ?if (-f $document_root/maintenance.php) { > ? ? ? ? ? ? ? ?rewrite ?^(.*)$ ?/maintenance.php last; > ? ? ? ? ? ? ? ?break; > ? ? ? ?} > } > > I'm not sure how to do the no maintenance part. > location ~ ^/(?!privacypolicy.php|images|promo) { try_files /maintenance.php $uri @fallback; } -- O< ascii ribbon campaign - stop html mail - www.asciiribbon.org From dave at cheney.net Thu Jul 2 09:43:16 2009 From: dave at cheney.net (Dave Cheney) Date: Thu, 02 Jul 2009 01:43:16 -0400 Subject: Tuning workers and connections References: <33c66c80907011823r7c6224e8w2945765c150f1f55@mail.gmail.com> Message-ID: Try adding events { ... use epoll; } http://wiki.nginx.org/NginxHttpEventsModule#use Cheers Dave Avleen Vig writes: > Hi folks, I have some questions about tuning nginx for best > performance, on a site which handle ~1000 requests per second. > > In our config, we have: > > worker_processes 16; > worker_rlimit_nofile 32768; > events { > worker_connections 8192; > } > > This is on a server with 8 CPU cores and 8Gb RAM. > My understanding is that this should allow nginx to establish up to > 32k network connections, because that is the limit of > worker_rlimit_nofile. We serve very few files off the disk, maybe a > dozen CSS and JS files. Everything else is handed to backends using > proxy_pass. > So with one fd for the browser, and one fd to the upstream, we should > be able to handle 16k concurrent connections. > > However what we're seeing, is that around 5k connections we get a > performance hit. Connections slow down, take longer to establish, etc. > The load on the box is almost zero, nginx is clearly working very fast > and efficiently, but I'm not sure why it slows down. > Any thoughts? > > > Also, I'm thinking about enabling the multi_accept option, but I > couldn't find much documentation on how this works. > It sounds like a high-performance tweak which we should be using if we > get this many requests per second. > Could it be that not using multi_accept is the problem? > Nginx otherwise has to handle the incoming connection requests in > serial and this is a bottleneck? > > > Thanks :) > From grails at jmsd.co.uk Thu Jul 2 13:07:58 2009 From: grails at jmsd.co.uk (John Moore) Date: Thu, 02 Jul 2009 10:07:58 +0100 Subject: Treat .htm files as PHP? In-Reply-To: <0FD4C1CE-C908-4365-A5B2-BA3A2C14327F@gmail.com> References: <4A4BCEB2.8010203@jmsd.co.uk> <0FD4C1CE-C908-4365-A5B2-BA3A2C14327F@gmail.com> Message-ID: <4A4C78EE.5020804@jmsd.co.uk> Michael Shadle wrote: > Should be as simple as changing a php block and changing the extension > to htm and passing it on to fastcgi like normal. I think. Hmm, I'm not sure I quite understand what you mean here. Could you perhaps give me an example? When I did what I think you meant, I got an error page saying 'No input file specified'. Thx. From mdounin at mdounin.ru Thu Jul 2 13:14:05 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Thu, 2 Jul 2009 13:14:05 +0400 Subject: Curious about old bug In-Reply-To: <33c66c80907011622ye551a2cna4abd7e936a06939@mail.gmail.com> References: <33c66c80907011622ye551a2cna4abd7e936a06939@mail.gmail.com> Message-ID: <20090702091405.GM5360@mdounin.ru> Hello! On Thu, Jul 02, 2009 at 12:22:22AM +0100, Avleen Vig wrote: > I'm reading through the changelog from 0.6.35 -> 0.7.59, and saw this: > Bugfix: now accept threshold depends on worker_connections > > I was wondering if anyone could tell me what that was? Historically nginx worker stops accepting new connections for a while if there are less than 100 connection structures available (to make sure that already accepted connections may finish it's work - since connection structures are used for proxy etc). In 0.7.18 this threshold was changed from constant one to worker_connections/8 to make sure that even workers configured with less than 100 workers_connections can accept new connections normally. This shouldn't affect typical configurations (worker_connections > 1024). Maxim Dounin From miles at tinyapps.org Thu Jul 2 13:13:33 2009 From: miles at tinyapps.org (TinyApps.Org) Date: Wed, 1 Jul 2009 23:13:33 -1000 Subject: Password-protected dirs working but reporting non-existent errors Message-ID: <58CF188B-D949-4A09-80DB-5663270DDDAF@tinyapps.org> A quick question if anyone can help: Password protected directories are working fine, but when a legitimate user logs in, the following appears in the error log: [error] 3876#0: *12 no user/password was provided for basic authentication despite the fact that the username and password *was* supplied. A bad username/password produces the expected error: [error] 3876#0: *12 user "whatever" was not found in "/etc/nginx/htpass" Has anyone else seen this or have any idea why such an error would be triggered? From phill at theactivitypeople.co.uk Thu Jul 2 13:28:20 2009 From: phill at theactivitypeople.co.uk (Phillip Oldham) Date: Thu, 02 Jul 2009 10:28:20 +0100 Subject: Treat .htm files as PHP? In-Reply-To: <4A4C78EE.5020804@jmsd.co.uk> References: <4A4BCEB2.8010203@jmsd.co.uk> <0FD4C1CE-C908-4365-A5B2-BA3A2C14327F@gmail.com> <4A4C78EE.5020804@jmsd.co.uk> Message-ID: <4A4C7DB4.7060603@theactivitypeople.co.uk> John Moore wrote: > Michael Shadle wrote: >> Should be as simple as changing a php block and changing the >> extension to htm and passing it on to fastcgi like normal. I think. > Hmm, I'm not sure I quite understand what you mean here. Could you > perhaps give me an example? When I did what I think you meant, I got > an error page saying 'No input file specified'. > > Thx. Follow this: http://wiki.nginx.org/NginxFcgiExample - and just change the location block so it checks for .html instead of .php. More details can be found here: http://wiki.nginx.org/NginxConfiguration http://wiki.nginx.org/NginxHttpFcgiModule From mdounin at mdounin.ru Thu Jul 2 13:56:00 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Thu, 2 Jul 2009 13:56:00 +0400 Subject: Tuning workers and connections In-Reply-To: <33c66c80907011823r7c6224e8w2945765c150f1f55@mail.gmail.com> References: <33c66c80907011823r7c6224e8w2945765c150f1f55@mail.gmail.com> Message-ID: <20090702095600.GN5360@mdounin.ru> Hello! On Thu, Jul 02, 2009 at 02:23:06AM +0100, Avleen Vig wrote: > Hi folks, I have some questions about tuning nginx for best > performance, on a site which handle ~1000 requests per second. > > In our config, we have: > > worker_processes 16; > worker_rlimit_nofile 32768; > events { > worker_connections 8192; > } > > This is on a server with 8 CPU cores and 8Gb RAM. > My understanding is that this should allow nginx to establish up to > 32k network connections, because that is the limit of > worker_rlimit_nofile. No. With this settings you will be able to handle up to 8192 connections in each worker, about 128k in total for 16 workers. If your system can cope with it. Directive worker_rlimit_nofile just raises open files limit for nginx processes if system allows this, nothing more. It's usually needed only if you started nginx with low limit, then raised the system one and want nginx to raise it's limit without restarting. > We serve very few files off the disk, maybe a > dozen CSS and JS files. Everything else is handed to backends using > proxy_pass. > So with one fd for the browser, and one fd to the upstream, we should > be able to handle 16k concurrent connections. > > However what we're seeing, is that around 5k connections we get a > performance hit. Connections slow down, take longer to establish, etc. > The load on the box is almost zero, nginx is clearly working very fast > and efficiently, but I'm not sure why it slows down. > Any thoughts? There are several things to check: 1. Which event method used? Make sure you use kqueue for *BSD, epoll for Linux. 2. What's in error_log? Your system logs/memory stats/etc? You may run out of some resources (file descriptors, network buffers, connection states in your firewall, ...). 3. In which states nginx workers are? If they are disk bound it's probably a good idea to check why and e.g. try to tune proxy buffers / output buffers / sendfile_max_chunk. > Also, I'm thinking about enabling the multi_accept option, but I > couldn't find much documentation on how this works. > It sounds like a high-performance tweak which we should be using if we > get this many requests per second. > Could it be that not using multi_accept is the problem? > Nginx otherwise has to handle the incoming connection requests in > serial and this is a bottleneck? With multi_accept nginx tries to accept all connections from listen queue at once. Without it - only one connection will be accepted on each return from event function. The bad thing with multi_accept that if you have constant stream of incomming connections at high rate - it may overflow your worker_connections without any chance to process already accepted connections. Note that this setting doesn't matter for kqueue (there is number of unaccepted connection at return from kernel, and nginx accepts all them). And it's always on for rtsig (not sure, but likely due to the fact that rtsig is fragile). Maxim Dounin From grails at jmsd.co.uk Thu Jul 2 14:05:48 2009 From: grails at jmsd.co.uk (John Moore) Date: Thu, 02 Jul 2009 11:05:48 +0100 Subject: Treat .htm files as PHP? In-Reply-To: <4A4C7DB4.7060603@theactivitypeople.co.uk> References: <4A4BCEB2.8010203@jmsd.co.uk> <0FD4C1CE-C908-4365-A5B2-BA3A2C14327F@gmail.com> <4A4C78EE.5020804@jmsd.co.uk> <4A4C7DB4.7060603@theactivitypeople.co.uk> Message-ID: <4A4C867C.5090702@jmsd.co.uk> Phillip Oldham wrote: > John Moore wrote: >> Michael Shadle wrote: >>> Should be as simple as changing a php block and changing the >>> extension to htm and passing it on to fastcgi like normal. I think. >> Hmm, I'm not sure I quite understand what you mean here. Could you >> perhaps give me an example? When I did what I think you meant, I got >> an error page saying 'No input file specified'. >> >> Thx. > Follow this: http://wiki.nginx.org/NginxFcgiExample - and just change > the location block so it checks for .html instead of .php. > > More details can be found here: > http://wiki.nginx.org/NginxConfiguration > http://wiki.nginx.org/NginxHttpFcgiModule > > > Thanks. My problem was that while PHP was working generally, I couldn't get it to work for the specific domain. I realized that I was misinforming FastCGI about the location, with this line: fastcgi_param SCRIPT_FILENAME /var/www/nginx-default$fastcgi_script_name; Changing it to this solved my problem: fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name; Thanks for the help. From nginx-forum at nginx.us Thu Jul 2 14:12:24 2009 From: nginx-forum at nginx.us (kaktos) Date: Thu, 2 Jul 2009 06:12:24 -0400 Subject: two nginx with ip_hash as a load balancer Message-ID: <5fbd449938953529977b3e714c12347c.NginxMailingList@forum.nginx.org> Hi, I have build a new application behind our hardware load balancer, since we need the 'sticky session' feather that our LB couldn't offer.So I put two nginx as a soft LB behind the hardware LB(that round-robin to nginx). Then I found the two nginx always hash the same client IP to different backend server. it's weird if I understood right about the ip_hash that using crc32 hashing the IP address and % the number of server nodes to get the target backend server. So why I got two different backend server? any help? Thank you! Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3660,3660#msg-3660 From nbubingo at gmail.com Thu Jul 2 14:34:43 2009 From: nbubingo at gmail.com (Weibin Yao) Date: Thu, 02 Jul 2009 18:34:43 +0800 Subject: two nginx with ip_hash as a load balancer In-Reply-To: <5fbd449938953529977b3e714c12347c.NginxMailingList@forum.nginx.org> References: <5fbd449938953529977b3e714c12347c.NginxMailingList@forum.nginx.org> Message-ID: <4A4C8D43.4030200@gmail.com> kaktos at 2009-7-2 18:12 wrote: > Hi, > I have build a new application behind our hardware load balancer, since we need the 'sticky session' feather that our LB couldn't offer.So I put two nginx as a soft LB behind the hardware LB(that round-robin to nginx). Then I found the two nginx > always hash the same client IP to different backend server. it's weird if I understood right about the ip_hash that using crc32 hashing the IP address and % the number of server nodes to get the target backend server. So why I got two different backend server? any help? > > You can try my developing module for session sticky: http://code.google.com/p/nginx-upstream-jvm-route/. And you need a careful test before using this module in your product environment. Questions/patches can be directed to me, hope this will help you. -- Weibin Yao From eugen at leitl.org Thu Jul 2 15:11:46 2009 From: eugen at leitl.org (Eugen Leitl) Date: Thu, 2 Jul 2009 13:11:46 +0200 Subject: moinmoin + nginx (+memcached) on debian? Message-ID: <20090702111146.GA22105@leitl.org> I've just jumped ship from lighty (lighttpd) and would like to configure nginx + moinmoin (fcgi, I presume) and maybe memcached on a Debian box (it's a Linux vserver, so no localhost). The nginx version from backported is 0.4.13 which is probably ancient. Do I need to upgrade? If I have to, does have someone to build a recipe for building a debian package for latest nginx tarball? Finally, has somebody a working config of nginx + moinmoin they can share? (Extre kudos for nginx + moinmoin + memcached). It looks like the snippet from http://johnleach.co.uk/words/archives/category/tech/ is not complete. I've added a server { listen 80; server_name cryoeuro.eu; access_log /var/log/nginx/localhost.access.log; location / { include /etc/nginx/fastcgi_params; fastcgi_param PATH_INFO $fastcgi_script_name; fastcgi_pass 88.198.119.230:9005; } ... if (!-f $request_filename) { fastcgi_pass 88.198.119.230:9005; } } } but I'm getting a 502 Bad Gateway nginx/0.5.35 Any ideas? Thanks. -- Eugen* Leitl leitl http://leitl.org ______________________________________________________________ ICBM: 48.07100, 11.36820 http://www.ativel.com http://postbiota.org 8B29F6BE: 099D 78BA 2FD3 B014 B08A 7779 75B0 2443 8B29 F6BE From admin at ampprod.com Thu Jul 2 16:52:42 2009 From: admin at ampprod.com (AMP Admin) Date: Thu, 2 Jul 2009 07:52:42 -0500 Subject: Advanced Maintenance In-Reply-To: References: <99e8cb160906302309q37c5b209o17fd68b1f52bbdb@mail.gmail.com> <20090701093047.GC5360@mdounin.ru> <4A4B935D.3030803@gmail.com> <013d01c9fa73$9c781bf0$d56853d0$@com> <20090701231921.GL5360@mdounin.ru> <024b01c9fac2$0e91b7c0$2bb52740$@com> Message-ID: <02d601c9fb13$fd96aaf0$f8c400d0$@com> Ohhh... i was looking for a not operator like '!'. thanks, I'll give that a try! -----Original Message----- From: owner-nginx at sysoev.ru [mailto:owner-nginx at sysoev.ru] On Behalf Of Edho P Arief Sent: Wednesday, July 01, 2009 10:38 PM To: nginx at sysoev.ru Subject: Re: Advanced Maintenance On Thu, Jul 2, 2009 at 10:06 AM, AMP Admin wrote: > I don't understand. How do we make the "no maintenance checks"? sorry, I'm > still learning. > > If we want everything but privacypolicy.php, images directory and promo > directory to go to maintenance how would that look? > > It seems that this puts the entire site in maintenance mode: > Location / { > if (-f $document_root/maintenance.php) { > rewrite ^(.*)$ /maintenance.php last; > break; > } > } > > I'm not sure how to do the no maintenance part. > location ~ ^/(?!privacypolicy.php|images|promo) { try_files /maintenance.php $uri @fallback; } -- O< ascii ribbon campaign - stop html mail - www.asciiribbon.org From avleen at gmail.com Thu Jul 2 18:01:13 2009 From: avleen at gmail.com (Avleen Vig) Date: Thu, 2 Jul 2009 15:01:13 +0100 Subject: Tuning workers and connections In-Reply-To: <20090702095600.GN5360@mdounin.ru> References: <33c66c80907011823r7c6224e8w2945765c150f1f55@mail.gmail.com> <20090702095600.GN5360@mdounin.ru> Message-ID: <33c66c80907020701n46f1683md6656384508ec72e@mail.gmail.com> > Directive worker_rlimit_nofile just raises open files limit for > nginx processes if system allows this, nothing more. ?It's usually > needed only if you started nginx with low limit, then raised the > system one and want nginx to raise it's limit without restarting. Ah thanks :) >> We serve very few files off the disk, maybe a >> dozen CSS and JS files. Everything else is handed to backends using >> proxy_pass. >> So with one fd for the browser, and one fd to the upstream, we should >> be able to handle 16k concurrent connections. >> >> However what we're seeing, is that around 5k connections we get a >> performance hit. Connections slow down, take longer to establish, etc. >> The load on the box is almost zero, nginx is clearly working very fast >> and efficiently, but I'm not sure why it slows down. >> Any thoughts? > > There are several things to check: > > 1. Which event method used? ?Make sure you use kqueue for *BSD, > epoll for Linux. epoll. > 2. What's in error_log? ?Your system logs/memory stats/etc? ?You > may run out of some resources (file descriptors, network buffers, > connection states in your firewall, ...). I saw nothing about resource starvation in the error log. Also I should mention this was on 0.6.35. We've also upgraded to 0.7.59. > 3. In which states nginx workers are? ?If they are disk bound it's > probably a good idea to check why and e.g. try to tune proxy > buffers / output buffers / sendfile_max_chunk. They don't appear to be disk bound. I will check again tonight at our peak hours. >> Also, I'm thinking about enabling the multi_accept option, but I >> couldn't find much documentation on how this works. >> It sounds like a high-performance tweak which we should be using if we >> get this many requests per second. >> Could it be that not using multi_accept is the problem? >> Nginx otherwise has to handle the incoming connection requests in >> serial and this is a bottleneck? > > With multi_accept nginx tries to accept all connections from > listen queue at once. ?Without it - only one connection will be > accepted on each return from event function. ?The bad thing with > multi_accept that if you have constant stream of incomming > connections at high rate - it may overflow your worker_connections > without any chance to process already accepted connections. Ok :-) I've massively increased the number of worker_connections now, and the box is otherwise quite powerful. It should be able to cope, but there's only one way to find out! Thanks guys! From avleen at gmail.com Thu Jul 2 17:58:11 2009 From: avleen at gmail.com (Avleen Vig) Date: Thu, 2 Jul 2009 14:58:11 +0100 Subject: Tuning workers and connections In-Reply-To: References: <33c66c80907011823r7c6224e8w2945765c150f1f55@mail.gmail.com> Message-ID: <33c66c80907020658q3b3cb8d5u481ecdeec052f661@mail.gmail.com> On Thu, Jul 2, 2009 at 6:43 AM, Dave Cheney wrote: > Try adding > > events { > ?... > ?use epoll; > } > > http://wiki.nginx.org/NginxHttpEventsModule#use Thanks Dave! Turns out this was the default (we're using a locally recompiled version of the Ubuntu package) :-) From mdounin at mdounin.ru Thu Jul 2 19:24:58 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Thu, 2 Jul 2009 19:24:58 +0400 Subject: Advanced Maintenance In-Reply-To: <02d601c9fb13$fd96aaf0$f8c400d0$@com> References: <99e8cb160906302309q37c5b209o17fd68b1f52bbdb@mail.gmail.com> <20090701093047.GC5360@mdounin.ru> <4A4B935D.3030803@gmail.com> <013d01c9fa73$9c781bf0$d56853d0$@com> <20090701231921.GL5360@mdounin.ru> <024b01c9fac2$0e91b7c0$2bb52740$@com> <02d601c9fb13$fd96aaf0$f8c400d0$@com> Message-ID: <20090702152458.GS5360@mdounin.ru> Hello! On Thu, Jul 02, 2009 at 07:52:42AM -0500, AMP Admin wrote: > Ohhh... i was looking for a not operator like '!'. thanks, I'll give that a try! It's much better to just configure normal more specific locations instead of using regexps, e.g. location / { # maintanance checks here } location /maintanance.php { # no checks here } location /privacypolicy.php { # and no checks here } location /images/ { # and no checks here } location /promo/ { # and no checks here } Each request will be processed in most-specific location, and hence there would be no maintanance checks for privacypolicy.php, images and promo (and maintanance.php itself). Maxim Dounin > > -----Original Message----- > From: owner-nginx at sysoev.ru [mailto:owner-nginx at sysoev.ru] On Behalf Of Edho P Arief > Sent: Wednesday, July 01, 2009 10:38 PM > To: nginx at sysoev.ru > Subject: Re: Advanced Maintenance > > On Thu, Jul 2, 2009 at 10:06 AM, AMP Admin wrote: > > I don't understand. How do we make the "no maintenance checks"? sorry, I'm > > still learning. > > > > If we want everything but privacypolicy.php, images directory and promo > > directory to go to maintenance how would that look? > > > > It seems that this puts the entire site in maintenance mode: > > Location / { > > if (-f $document_root/maintenance.php) { > > rewrite ^(.*)$ /maintenance.php last; > > break; > > } > > } > > > > I'm not sure how to do the no maintenance part. > > > > location ~ ^/(?!privacypolicy.php|images|promo) { > try_files /maintenance.php $uri @fallback; > } > > -- > O< ascii ribbon campaign - stop html mail - www.asciiribbon.org > > From nginx-forum at nginx.us Thu Jul 2 19:29:07 2009 From: nginx-forum at nginx.us (miradev) Date: Thu, 2 Jul 2009 11:29:07 -0400 Subject: More subdomain mapping (dynamic 'root'?) In-Reply-To: References: Message-ID: No replies as yet. This is what I have come up with so, but I'm not sure I am passing on the correct values to fcgi. And I would like the log file locations to be a little dynamic. server { listen localhost:80; # your server's public IP address server_name .admin.domain.com; access_log /var/log/nginx/wordpress/admin/generic-access.log; error_log /var/log/nginx/wordpress/admin/generic-error.log; set $client $host; if ($client ~ "^(*).admin.domain.com"){ set $client $1; } location ~ ^/([^/]*)/?(.*) { set $domain $1; set $myuri $2; root /var/www/wordpress/$client/admin/$domain/html; index index.php index.html index.htm; if (-f $request_filename) { expires 30d; break; } try_files $domain/$myuri $domain/$myuri/ /$domain/index.php?q=$myuri&$args; } } location ~ \.php$ { fastcgi_pass localhost:57093; # port where FastCGI processes were spawned fastcgi_index index.php; fastcgi_param SCRIPT_FILENAME /var/www/wordpress/$client/admin/$domain/html$fastcgi_script_name; # same path as above fastcgi_param QUERY_STRING $query_string; fastcgi_param REQUEST_METHOD $request_method; fastcgi_param CONTENT_TYPE $content_type; fastcgi_param CONTENT_LENGTH $content_length; fastcgi_param SCRIPT_NAME $fastcgi_script_name; fastcgi_param REQUEST_URI $request_uri; fastcgi_param DOCUMENT_URI $document_uri; fastcgi_param DOCUMENT_ROOT $document_root; fastcgi_param SERVER_PROTOCOL $server_protocol; fastcgi_param GATEWAY_INTERFACE CGI/1.1; fastcgi_param SERVER_SOFTWARE nginx/$nginx_version; fastcgi_param REMOTE_ADDR $remote_addr; fastcgi_param REMOTE_PORT $remote_port; fastcgi_param SERVER_ADDR $server_addr; fastcgi_param SERVER_PORT $server_port; fastcgi_param SERVER_NAME $server_name; # required if PHP was built with --enable-force-cgi-redirect fastcgi_param REDIRECT_STATUS 200; } } Any helpful comments or musings much appreciated. Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3586,3673#msg-3673 From mike503 at gmail.com Thu Jul 2 21:23:16 2009 From: mike503 at gmail.com (Michael Shadle) Date: Thu, 2 Jul 2009 10:23:16 -0700 Subject: Treat .htm files as PHP? In-Reply-To: <4A4C867C.5090702@jmsd.co.uk> References: <4A4BCEB2.8010203@jmsd.co.uk> <0FD4C1CE-C908-4365-A5B2-BA3A2C14327F@gmail.com> <4A4C78EE.5020804@jmsd.co.uk> <4A4C7DB4.7060603@theactivitypeople.co.uk> <4A4C867C.5090702@jmsd.co.uk> Message-ID: That's how I have mine setup. No need to hardcode the root in twice! Sent from my iPhone On Jul 2, 2009, at 3:05 AM, John Moore wrote: > Phillip Oldham wrote: >> John Moore wrote: >>> Michael Shadle wrote: >>>> Should be as simple as changing a php block and changing the >>>> extension to htm and passing it on to fastcgi like normal. I think. >>> Hmm, I'm not sure I quite understand what you mean here. Could you >>> perhaps give me an example? When I did what I think you meant, I >>> got an error page saying 'No input file specified'. >>> >>> Thx. >> Follow this: http://wiki.nginx.org/NginxFcgiExample - and just >> change the location block so it checks for .html instead of .php. >> >> More details can be found here: >> http://wiki.nginx.org/NginxConfiguration >> http://wiki.nginx.org/NginxHttpFcgiModule >> >> >> > Thanks. My problem was that while PHP was working generally, I > couldn't get it to work for the specific domain. I realized that I > was misinforming FastCGI about the location, with this line: > > fastcgi_param SCRIPT_FILENAME /var/www/nginx-default > $fastcgi_script_name; > > Changing it to this solved my problem: > > fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name; > > Thanks for the help. > From mike503 at gmail.com Thu Jul 2 21:56:26 2009 From: mike503 at gmail.com (Michael Shadle) Date: Thu, 2 Jul 2009 10:56:26 -0700 Subject: Tuning workers and connections In-Reply-To: <20090702095600.GN5360@mdounin.ru> References: <33c66c80907011823r7c6224e8w2945765c150f1f55@mail.gmail.com> <20090702095600.GN5360@mdounin.ru> Message-ID: On Thu, Jul 2, 2009 at 2:56 AM, Maxim Dounin wrote: > 1. Which event method used? ?Make sure you use kqueue for *BSD, > epoll for Linux. Is epoll enabled by default? Does it make sense to enable this for all servers, or only those with massive amounts of load? From yazz.atlas at gmail.com Thu Jul 2 22:01:40 2009 From: yazz.atlas at gmail.com (Yazz Atlas) Date: Thu, 02 Jul 2009 11:01:40 -0700 Subject: Help with autoindex on; missing slash for directories In-Reply-To: <20090701225352.GJ5360@mdounin.ru> References: <99e8cb160906302309q37c5b209o17fd68b1f52bbdb@mail.gmail.com> <20090701093047.GC5360@mdounin.ru> <4A4B935D.3030803@gmail.com> <20090701225352.GJ5360@mdounin.ru> Message-ID: <4A4CF604.5030508@gmail.com> On 7/1/09 3:53 PM, Maxim Dounin wrote: > > I mean *appeared* in 0.7.40, i.e. problem exists in 0.7.40 but not > in 0.7.39. But after some thinking it's most likely 0.7.15 when > dirent.d_type checking was initially introduced (i.e. problem exists in > 0.7.15 but not in 0.7.14). Ah, gotcha. I miss understood. > > [...] > >> checking for struct dirent.d_namlen ... not found >> checking for struct dirent.d_type ... found > > [...] > > You system has dirent.d_type, but it looks like it's something > wrong with it. I've tried to reproduce problem on vmware image of > CentOS 5.3 i386 - but failed, everythings works ok. I've even > tried initial release version and fully updated one (with > 2.6.18-128.1.16.el5 kernel), no problems found. > > It's either x64 only issue, or something wrong with headers on > your server(s). Unfortenately I have no x64 version on hand to > test it there, but people report that everything is ok at least on > other x64 distributions. > I tested under VMware Server Version 2.0.1 and selecting Red Hat Enterprise Linux (64bit) as the Guest Operating System. Then I loaded CentOS release 5.3 (Final) x86_64 with Kernel 2.6.18-128.1.14.el5. The problem still exists. > Could you please check your headers, most notably > /usr/include/dirent.h and /usr/include/bits/dirent.h? > I attached them in the file called include-dirent-x86_64.tar.gz. include/dirent.h include/bits/dirent.h Yazz Atlas -------------- next part -------------- A non-text attachment was scrubbed... Name: include-dirent-x86_64.tar.gz Type: application/x-gzip Size: 3704 bytes Desc: not available URL: From mdounin at mdounin.ru Thu Jul 2 22:19:29 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Thu, 2 Jul 2009 22:19:29 +0400 Subject: Tuning workers and connections In-Reply-To: References: <33c66c80907011823r7c6224e8w2945765c150f1f55@mail.gmail.com> <20090702095600.GN5360@mdounin.ru> Message-ID: <20090702181929.GV5360@mdounin.ru> Hello! On Thu, Jul 02, 2009 at 10:56:26AM -0700, Michael Shadle wrote: > On Thu, Jul 2, 2009 at 2:56 AM, Maxim Dounin wrote: > > > 1. Which event method used? ?Make sure you use kqueue for *BSD, > > epoll for Linux. > > Is epoll enabled by default? By default nginx uses best event method available for particular system. If it finds epoll during ./configure - it uses epoll. > Does it make sense to enable this for all servers, or only those with > massive amounts of load? There is no reason to use slow/limited event methods if better ones are available. Maxim Dounin From mike503 at gmail.com Thu Jul 2 22:34:43 2009 From: mike503 at gmail.com (Michael Shadle) Date: Thu, 2 Jul 2009 11:34:43 -0700 Subject: Tuning workers and connections In-Reply-To: <20090702181929.GV5360@mdounin.ru> References: <33c66c80907011823r7c6224e8w2945765c150f1f55@mail.gmail.com> <20090702095600.GN5360@mdounin.ru> <20090702181929.GV5360@mdounin.ru> Message-ID: 2009/7/2 Maxim Dounin : > Hello! > > On Thu, Jul 02, 2009 at 10:56:26AM -0700, Michael Shadle wrote: > >> On Thu, Jul 2, 2009 at 2:56 AM, Maxim Dounin wrote: >> >> > 1. Which event method used? ?Make sure you use kqueue for *BSD, >> > epoll for Linux. >> >> Is epoll enabled by default? > > By default nginx uses best event method available for particular > system. ?If it finds epoll during ./configure - it uses epoll. That's what I thought. >> Does it make sense to enable this for all servers, or only those with >> massive amounts of load? > > There is no reason to use slow/limited event methods if better > ones are available. How can I tell which method I'm currently using? From mdounin at mdounin.ru Thu Jul 2 23:05:21 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Thu, 2 Jul 2009 23:05:21 +0400 Subject: Tuning workers and connections In-Reply-To: References: <33c66c80907011823r7c6224e8w2945765c150f1f55@mail.gmail.com> <20090702095600.GN5360@mdounin.ru> <20090702181929.GV5360@mdounin.ru> Message-ID: <20090702190521.GW5360@mdounin.ru> Hello! On Thu, Jul 02, 2009 at 11:34:43AM -0700, Michael Shadle wrote: [...] > >> Does it make sense to enable this for all servers, or only those with > >> massive amounts of load? > > > > There is no reason to use slow/limited event methods if better > > ones are available. > > How can I tell which method I'm currently using? At start nginx logs it at notice level, e.g. 2009/07/02 23:04:02 [notice] 83443#0: using the "kqueue" event method Maxim Dounin From mdounin at mdounin.ru Thu Jul 2 23:22:52 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Thu, 2 Jul 2009 23:22:52 +0400 Subject: Help with autoindex on; missing slash for directories In-Reply-To: <4A4CF604.5030508@gmail.com> References: <99e8cb160906302309q37c5b209o17fd68b1f52bbdb@mail.gmail.com> <20090701093047.GC5360@mdounin.ru> <4A4B935D.3030803@gmail.com> <20090701225352.GJ5360@mdounin.ru> <4A4CF604.5030508@gmail.com> Message-ID: <20090702192251.GX5360@mdounin.ru> Hello! On Thu, Jul 02, 2009 at 11:01:40AM -0700, Yazz Atlas wrote: > On 7/1/09 3:53 PM, Maxim Dounin wrote: >> >> I mean *appeared* in 0.7.40, i.e. problem exists in 0.7.40 but not >> in 0.7.39. But after some thinking it's most likely 0.7.15 when >> dirent.d_type checking was initially introduced (i.e. problem exists in >> 0.7.15 but not in 0.7.14). > > Ah, gotcha. I miss understood. > >> >> [...] >> >>> checking for struct dirent.d_namlen ... not found >>> checking for struct dirent.d_type ... found >> >> [...] >> >> You system has dirent.d_type, but it looks like it's something >> wrong with it. I've tried to reproduce problem on vmware image of >> CentOS 5.3 i386 - but failed, everythings works ok. I've even >> tried initial release version and fully updated one (with >> 2.6.18-128.1.16.el5 kernel), no problems found. >> >> It's either x64 only issue, or something wrong with headers on >> your server(s). Unfortenately I have no x64 version on hand to >> test it there, but people report that everything is ok at least on >> other x64 distributions. >> > > I tested under VMware Server Version 2.0.1 and selecting Red Hat > Enterprise Linux (64bit) as the Guest Operating System. Then I loaded > CentOS release 5.3 (Final) x86_64 with Kernel 2.6.18-128.1.14.el5. > > The problem still exists. > >> Could you please check your headers, most notably >> /usr/include/dirent.h and /usr/include/bits/dirent.h? >> > > I attached them in the file called include-dirent-x86_64.tar.gz. > > include/dirent.h > include/bits/dirent.h Looks good, I see nothing unusual there. Could you please compile nginx with the attached simple patch and obtain debug log for autoindex request? Maxim Dounin -------------- next part -------------- diff -r 1191e3250e60 src/http/modules/ngx_http_autoindex_module.c --- a/src/http/modules/ngx_http_autoindex_module.c Mon Jun 22 00:00:00 2009 +0400 +++ b/src/http/modules/ngx_http_autoindex_module.c Thu Jul 02 19:22:42 2009 +0000 @@ -278,7 +278,7 @@ } ngx_log_debug1(NGX_LOG_DEBUG_HTTP, r->connection->log, 0, - "http autoindex file: \"%s\"", ngx_de_name(&dir)); + "http autoindex file: \"%s\", type %d", ngx_de_name(&dir), dir.type); len = ngx_de_namelen(&dir); From nunomagalhaes at eu.ipp.pt Fri Jul 3 01:08:37 2009 From: nunomagalhaes at eu.ipp.pt (=?UTF-8?Q?Nuno_Magalh=C3=A3es?=) Date: Thu, 2 Jul 2009 22:08:37 +0100 Subject: server_name ip_adress? Message-ID: <6b1504c40907021408u166d9ca0h15b4ad8de3fe3f30@mail.gmail.com> Hi, This is a very basic question i haven't found an answer to. I have a few virtual hosts, all under /path/to/nginxhtdocs/vhost1, vhost2, etc. I want to set a catch-all so that if no server matches the request or is someone uses the IP (namely me) they get the default server. After a bunch of variations i'm using server_name _ ""; with a server root of /path/to/nginxhtdocs but if i use the IP address to access it, i get one of my vhosts. I assume it does this because it fails to assume 'default' as a valid catch-all and the vhost it uses is the first, alphabetically. All i really want is to have other subfolders i want to access every now and then, without a domain, so, just 192.0.2.33/somesubfolder for testing purposes. Is this achievable or do i always need a domain? TIA, Nuno Magalh?es -- () ascii ribbon campaign - against html e-mail /\ ascii-rubanda kampajno - kontra? html-a retpo?to From nunomagalhaes at eu.ipp.pt Fri Jul 3 01:40:59 2009 From: nunomagalhaes at eu.ipp.pt (=?UTF-8?Q?Nuno_Magalh=C3=A3es?=) Date: Thu, 2 Jul 2009 22:40:59 +0100 Subject: server_name ip_adress? In-Reply-To: <6b1504c40907021408u166d9ca0h15b4ad8de3fe3f30@mail.gmail.com> References: <6b1504c40907021408u166d9ca0h15b4ad8de3fe3f30@mail.gmail.com> Message-ID: <6b1504c40907021440h282eeeacweccc1e9c9977e52f@mail.gmail.com> Got it. server { listen 80 default; server_name _ ""; "default" was lacking. I was considering naming my default to aaaadefault but knew there had to be an elegant way. Sorry for the noise. -- () ascii ribbon campaign - against html e-mail /\ ascii-rubanda kampajno - kontra? html-a retpo?to From zhuzhaoyuan at gmail.com Fri Jul 3 07:55:22 2009 From: zhuzhaoyuan at gmail.com (Joshua Zhu) Date: Fri, 3 Jul 2009 11:55:22 +0800 Subject: Help with autoindex on; missing slash for directories In-Reply-To: <20090702192251.GX5360@mdounin.ru> References: <99e8cb160906302309q37c5b209o17fd68b1f52bbdb@mail.gmail.com> <20090701093047.GC5360@mdounin.ru> <4A4B935D.3030803@gmail.com> <20090701225352.GJ5360@mdounin.ru> <4A4CF604.5030508@gmail.com> <20090702192251.GX5360@mdounin.ru> Message-ID: <5bf4451d0907022055y33964155m9d49df06b2a3e9ed@mail.gmail.com> Hi Maxim! > Could you please compile nginx with the attached simple patch and > obtain debug log for autoindex request? > IMHO, the ngx_log_debug1 in your patch should be changed to ngx_log_debug2 :) Cheers! Joshua From catap+nginx at catap.ru Fri Jul 3 10:59:53 2009 From: catap+nginx at catap.ru (Kirill A. Korinskiy) Date: Fri, 03 Jul 2009 10:59:53 +0400 Subject: server_name ip_adress? In-Reply-To: <6b1504c40907021440h282eeeacweccc1e9c9977e52f@mail.gmail.com> References: <6b1504c40907021408u166d9ca0h15b4ad8de3fe3f30@mail.gmail.com> <6b1504c40907021440h282eeeacweccc1e9c9977e52f@mail.gmail.com> Message-ID: <87ab3mtfo6.wl%catap@catap.ru> At Thu, 2 Jul 2009 22:40:59 +0100, Nuno Magalh?es wrote: > server { > listen 80 default; > server_name _ ""; > > "default" was lacking. I was considering naming my default to > aaaadefault but knew there had to be an elegant way. > or remove server_name derective (: -- wbr, Kirill From nginx-forum at nginx.us Fri Jul 3 11:03:54 2009 From: nginx-forum at nginx.us (afen) Date: Fri, 3 Jul 2009 03:03:54 -0400 Subject: how much Active connections on nginx for windows? In-Reply-To: References: Message-ID: what's wrong my config Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3233,3706#msg-3706 From mdounin at mdounin.ru Fri Jul 3 12:49:35 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Fri, 3 Jul 2009 12:49:35 +0400 Subject: Help with autoindex on; missing slash for directories In-Reply-To: <5bf4451d0907022055y33964155m9d49df06b2a3e9ed@mail.gmail.com> References: <99e8cb160906302309q37c5b209o17fd68b1f52bbdb@mail.gmail.com> <20090701093047.GC5360@mdounin.ru> <4A4B935D.3030803@gmail.com> <20090701225352.GJ5360@mdounin.ru> <4A4CF604.5030508@gmail.com> <20090702192251.GX5360@mdounin.ru> <5bf4451d0907022055y33964155m9d49df06b2a3e9ed@mail.gmail.com> Message-ID: <20090703084935.GY5360@mdounin.ru> Hello! On Fri, Jul 03, 2009 at 11:55:22AM +0800, Joshua Zhu wrote: > Hi Maxim! > > > Could you please compile nginx with the attached simple patch and > > obtain debug log for autoindex request? > > > > IMHO, the ngx_log_debug1 in your patch should be changed to ngx_log_debug2 :) Yes, you are right, I missed it. But it doesn't matter as long as nginx compiled --with-debug. And this in turn is required since I need this patch just for more verbose debug log on particular system. :) Maxim Dounin From slogster at gmail.com Fri Jul 3 12:58:08 2009 From: slogster at gmail.com (Momchil Ivanov) Date: Fri, 3 Jul 2009 10:58:08 +0200 Subject: Tuning workers and connections In-Reply-To: <33c66c80907011823r7c6224e8w2945765c150f1f55@mail.gmail.com> References: <33c66c80907011823r7c6224e8w2945765c150f1f55@mail.gmail.com> Message-ID: <200907031058.16034.slogster@gmail.com> ?? Thursday 02 July 2009 03:23:06 Avleen Vig ??????: > Hi folks, I have some questions about tuning nginx for best > performance, on a site which handle ~1000 requests per second. > > In our config, we have: > > worker_processes 16; > worker_rlimit_nofile 32768; > events { > worker_connections 8192; > } > > This is on a server with 8 CPU cores and 8Gb RAM. > My understanding is that this should allow nginx to establish up to > 32k network connections, because that is the limit of > worker_rlimit_nofile. We serve very few files off the disk, maybe a > dozen CSS and JS files. Everything else is handed to backends using > proxy_pass. > So with one fd for the browser, and one fd to the upstream, we should > be able to handle 16k concurrent connections. > > However what we're seeing, is that around 5k connections we get a > performance hit. Connections slow down, take longer to establish, etc. > The load on the box is almost zero, nginx is clearly working very fast > and efficiently, but I'm not sure why it slows down. > Any thoughts? > > > Also, I'm thinking about enabling the multi_accept option, but I > couldn't find much documentation on how this works. > It sounds like a high-performance tweak which we should be using if we > get this many requests per second. > Could it be that not using multi_accept is the problem? > Nginx otherwise has to handle the incoming connection requests in > serial and this is a bottleneck? The first thing to do would be playing with sysctls, especially increasing backlog and somaxconn, decreasing tcp_fin_timeout. If this doesn't help a lot you may consider using keepalive connections (only if you are hitting some issue with handling lots of accepts). If you move to keepalive connections, you should make sure that you befefit at all, therefore monitor requests/handled accepts. Momchil -- PGP KeyID: 0x3118168B Keyserver: pgp.mit.edu Key fingerprint BB50 2983 0714 36DC D02E ?158A E03D 56DA 3118 168B -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 198 bytes Desc: This is a digitally signed message part. URL: From frank at openminds.be Fri Jul 3 13:36:24 2009 From: frank at openminds.be (Frank Louwers) Date: Fri, 3 Jul 2009 11:36:24 +0200 Subject: More subdomain mapping (dynamic 'root'?) In-Reply-To: References: Message-ID: <4E19A7D2-7EFE-4D9E-BC2A-322FEA0EDEB7@openminds.be> On 01 Jul 2009, at 12:23, miradev wrote: > I want to be able to map submains and urls to determined locations > on disk, if they exist. > > The URL structure is something like: http://.admin.domain.com/ > > Where would letters, numbers and underscores , and is a domain > name (e.g. http://cnn.admin.domain.com/www.cnnminiblog.com) > This is what we do for a customer: this allows them to create new domains/projects just by making a new directory. With passenger and php-fpm and this setup, he doesn't need to restart/reload *anything* to start a new website. server { listen 80; server_name _; server_name_in_redirect off; passenger_enabled on; passenger_use_global_queue on; root /home/customer/sites/$host/public; if (!-d $document_root) { #site not (yet) configureerd. Serve default rewrite (.*) http://default.customer.tld/; } location / { passenger_enabled on; index index.html index.htm index.php; } location ~ .php$ { fastcgi_pass 127.0.0.1:9000; fastcgi_index index.php; include fastcgi_params; } } Regards, Frank -------------- next part -------------- An HTML attachment was scrubbed... URL: From nginx-forum at nginx.us Fri Jul 3 14:42:14 2009 From: nginx-forum at nginx.us (miradev) Date: Fri, 3 Jul 2009 06:42:14 -0400 Subject: More subdomain mapping (dynamic 'root'?) In-Reply-To: <4E19A7D2-7EFE-4D9E-BC2A-322FEA0EDEB7@forum.nginx.org> References: <4E19A7D2-7EFE-4D9E-BC2A-322FEA0EDEB7@forum.nginx.org> Message-ID: Frank Louwers Wrote: ------------------------------------------------------- > > This is what we do for a customer: this allows > them to create new > domains/projects just by making a new directory. > With passenger and > php-fpm and this setup, he doesn't need to > restart/reload *anything* > to start a new website. > > > server { > listen 80; > server_name _; > server_name_in_redirect off; > passenger_enabled on; > passenger_use_global_queue on; > root > /home/customer/sites/$host/public; > > if (!-d $document_root) { > #site not (yet) configureerd. > Serve default > rewrite (.*) > http://default.customer.tld/; > } > > location / { > passenger_enabled on; > index index.html index.htm > index.php; > } > location ~ .php$ { > fastcgi_pass 127.0.0.1:9000; > fastcgi_index index.php; > include fastcgi_params; > } > } > You are building one config file per client - you have '/home/customer/sites/' hard coded? I'm confused, sorry. With my example (http://cnn.admin.domain.com/www.cnnminiblog.com) I would end up in /home/customer/sites/cnn.admin.domain.com/public/www.cnnminiblog.com I'd like the public folder inside the domain, so like this: /home/cnn/sites/www.cnnminiblog.com/public Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3586,3720#msg-3720 From frank at openminds.be Fri Jul 3 16:56:42 2009 From: frank at openminds.be (Frank Louwers) Date: Fri, 3 Jul 2009 14:56:42 +0200 Subject: More subdomain mapping (dynamic 'root'?) In-Reply-To: References: <4E19A7D2-7EFE-4D9E-BC2A-322FEA0EDEB7@forum.nginx.org> Message-ID: On 03 Jul 2009, at 12:42, miradev wrote: >> > > You are building one config file per client - you have '/home/ > customer/sites/' hard coded? I'm confused, sorry. Yes, it's a dedicated server for a customer. > > With my example (http://cnn.admin.domain.com/www.cnnminiblog.com) I > would end up in > /home/customer/sites/cnn.admin.domain.com/public/www.cnnminiblog.com > > I'd like the public folder inside the domain, so like this: > /home/cnn/sites/www.cnnminiblog.com/public Can you give an example of two urls and the docroot you'd want them to use? Frank -------------- next part -------------- An HTML attachment was scrubbed... URL: From nginx-forum at nginx.us Fri Jul 3 17:28:11 2009 From: nginx-forum at nginx.us (miradev) Date: Fri, 3 Jul 2009 09:28:11 -0400 Subject: Variables in access_log Message-ID: <911f7c981591b165a63cb62d3117b09a.NginxMailingList@forum.nginx.org> >From http://wiki.nginx.org/NginxHttpLogModule The log file path can contain variables (version >=0.7.4) .. nginx -v nginx version: nginx/0.7.59 Config snippet: set $client $host; if ($client ~ "^(*).admin.mysite.com"){ set $client $1; } root /var/www/wordpress/$client/admin; location / { ... access_log /var/log/nginx/wordpress/admin/$client/generic-access.log; error_log /var/log/nginx/wordpress/admin/$client/generic-error.log; } Fails on test: nginx -t the configuration file /usr/local/nginx/conf/nginx.conf syntax is ok : open() "/var/log/nginx/wordpress/admin/$client/generic-error.log" failed (2: No such file or directory) configuration file /usr/local/nginx/conf/nginx.conf test failed Have I misunderstood? Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3722,3722#msg-3722 From eugen at leitl.org Fri Jul 3 17:42:38 2009 From: eugen at leitl.org (Eugen Leitl) Date: Fri, 3 Jul 2009 15:42:38 +0200 Subject: can someone post their working nginx+moinmoin config? Message-ID: <20090703134238.GA31310@leitl.org> Can someone post their working nginx+moinmoin config? Thanks/???????. -- Eugen* Leitl leitl http://leitl.org ______________________________________________________________ ICBM: 48.07100, 11.36820 http://www.ativel.com http://postbiota.org 8B29F6BE: 099D 78BA 2FD3 B014 B08A 7779 75B0 2443 8B29 F6BE From nginx-forum at nginx.us Fri Jul 3 18:19:47 2009 From: nginx-forum at nginx.us (miradev) Date: Fri, 3 Jul 2009 10:19:47 -0400 Subject: More subdomain mapping (dynamic 'root'?) In-Reply-To: References: <4E19A7D2-7EFE-4D9E-BC2A-322FEA0EDEB7@forum.nginx.org>

Message-ID: <9756bf68172578da057860103afc76a5.NginxMailingList@forum.nginx.org> Frank Louwers Wrote: ------------------------------------------------------- > Can you give an example of two urls and the > docroot you'd want them to > use? These are test/dev sites where clients view sample themes, copy and plugins. What they want is to be able to do is see them in action without affecting their 'live' site.. So these are all hosted on a single 'dev' domain. Admin domain: wpdev.mysite.com Client: ExoGarage Domain1: www.exogarage.com Dmoain2: www.fixurcar.com URL1: exogarage.wpdev.mysite.com/www.exogarage.com Docroot: /var/www/wordpress/clients/exogarage/admin/www.exogarage.com/html URL2: exogarage.wpdev.mysite.com/www.fixurcar.com Docroot: /var/www/wordpress/clients/exogarage/admin/www.fixurcar.com/html Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3586,3725#msg-3725 From cherife at dotimes.com Fri Jul 3 18:20:21 2009 From: cherife at dotimes.com (Cherife Li) Date: Fri, 3 Jul 2009 22:20:21 +0800 Subject: can someone post their working nginx+moinmoin config? In-Reply-To: <20090703134238.GA31310@leitl.org> References: <20090703134238.GA31310@leitl.org> Message-ID: <20090703142021.GA3505@dotimes.com> On Fri, 03 Jul 2009 at 15:42:38 +0200, Eugen Leitl wrote: > > Can someone post their working nginx+moinmoin config? > Thanks/???????. I'm running Moin standalone mode, and using proxy_redirect within nginx. Here is the block of the config: [..snip..] # For MoinMoin: rewrite ^/moin_static[0-9]+/(.*)$ /moin/$1 last; location /wiki/ { proxy_redirect http://localhost:8080 /; proxy_pass http://localhost:8080; proxy_set_header X-Real-IP $remote_addr; proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for; proxy_set_header Host $http_host; proxy_set_header X-Moin-Location /wiki/; } [..snip..] Cliff Wells has the experience using Twisted server. It's said to be better. Although I haven't tried it out. :-) -- Rgds, Cherife. -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 197 bytes Desc: not available URL: From brentb at beanfield.com Fri Jul 3 18:43:36 2009 From: brentb at beanfield.com (Brent Bloxam) Date: Fri, 03 Jul 2009 10:43:36 -0400 Subject: Variables in access_log In-Reply-To: <911f7c981591b165a63cb62d3117b09a.NginxMailingList@forum.nginx.org> References: <911f7c981591b165a63cb62d3117b09a.NginxMailingList@forum.nginx.org> Message-ID: <4A4E1918.9050909@beanfield.com> miradev wrote: > Fails on test: > > nginx -t > the configuration file /usr/local/nginx/conf/nginx.conf syntax is ok > : open() "/var/log/nginx/wordpress/admin/$client/generic-error.log" failed (2: No such file or directory) > configuration file /usr/local/nginx/conf/nginx.conf test failed > > > Have I misunderstood? > > Yes, you assumed error_log was the same as access_log. They are not. access_log is part of the Log module while error_log is part of the main module. error_log (at least in 0.7.59) does not support variables in the path. From nginx-forum at nginx.us Fri Jul 3 19:00:39 2009 From: nginx-forum at nginx.us (tmiskiew) Date: Fri, 3 Jul 2009 11:00:39 -0400 Subject: How to set Retry-After for 503? Message-ID: <418c2d7dfb3f50f6248f078e1a3c2620.NginxMailingList@forum.nginx.org> Hi everyone, we're using nginx/0.5.35 I want to set the Retry-After header for 503, but it isn't working. This is how I tried: error_page 500 502 504 /500.html; error_page 503 /system/maintenance.html; location /system/maintenance.html { # Allow requests } location / { if (-f $document_root/system/maintenance.html) { add_header Retry-After 600; return 503; } ... } I've checked with curl -I -H and it's not showing Retry-After. Any ideas? Thanks, Thomas Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3727,3727#msg-3727 From eugen at leitl.org Fri Jul 3 19:26:04 2009 From: eugen at leitl.org (Eugen Leitl) Date: Fri, 3 Jul 2009 17:26:04 +0200 Subject: can someone post their working nginx+moinmoin config? In-Reply-To: <20090703142021.GA3505@dotimes.com> References: <20090703134238.GA31310@leitl.org> <20090703142021.GA3505@dotimes.com> Message-ID: <20090703152604.GC31310@leitl.org> On Fri, Jul 03, 2009 at 10:20:21PM +0800, Cherife Li wrote: > Cliff Wells has the experience using Twisted server. It's said to be > better. Although I haven't tried it out. :-) Thank you. Have you considered using fcgi for MoinMoin, and if yes, did you reject it for a reason? -- Eugen* Leitl leitl http://leitl.org ______________________________________________________________ ICBM: 48.07100, 11.36820 http://www.ativel.com http://postbiota.org 8B29F6BE: 099D 78BA 2FD3 B014 B08A 7779 75B0 2443 8B29 F6BE From gmm at csdoc.com Fri Jul 3 20:13:59 2009 From: gmm at csdoc.com (Gena Makhomed) Date: Fri, 3 Jul 2009 19:13:59 +0300 Subject: How to set Retry-After for 503? In-Reply-To: <418c2d7dfb3f50f6248f078e1a3c2620.NginxMailingList@forum.nginx.org> References: <418c2d7dfb3f50f6248f078e1a3c2620.NginxMailingList@forum.nginx.org> Message-ID: <1483953504.20090703191359@csdoc.com> On Friday, July 3, 2009 at 18:00:39, tmiskiew wrote: t> we're using nginx/0.5.35 t> I want to set the Retry-After header for 503, but it isn't working. t> This is how I tried: t> location / { t> if (-f $document_root/system/maintenance.html) { t> add_header Retry-After 600; t> return 503; t> } t> I've checked with curl -I -H and it's not showing Retry-After. t> Any ideas? add_header works only for 200, 204, 301, 302 or 304 status codes. documentation: http://sysoev.ru/nginx/docs/http/ngx_http_headers_module.html -- Best regards, Gena From pedro.matamouros at sapo.pt Fri Jul 3 20:45:16 2009 From: pedro.matamouros at sapo.pt (Pedro Mata-Mouros Fonseca) Date: Fri, 3 Jul 2009 17:45:16 +0100 Subject: External health checking - OPTIONS returns 405 Message-ID: <10F8B021-CD78-450B-860F-0B0B8612E994@sapo.pt> Greetings, The following is an HAProxy option httpchk check request. It's purpose is to check a server's health by means of issuing a very lightweight request to it. It does this, by default, using OPTIONS (it's possible to configure it to perform any type of request, but OPTIONS is aparently the most lightweight). Nginx unfortunately returns a 405 to this. It's not really a big deal, but is there anyway to "force" Nginx to return a 200 with an empty payload or as small as possible? Would you still consider this a lightweight check to perform every 2 or so seconds, on a few hundred reqs/sec website? 127.0.0.1 - - [03/Jul/2009:17:34:21 +0100] "OPTIONS / HTTP/1.0" 405 166 "-" "-" Thanks for your time. Pedro. From mike503 at gmail.com Fri Jul 3 21:54:53 2009 From: mike503 at gmail.com (Michael Shadle) Date: Fri, 3 Jul 2009 10:54:53 -0700 Subject: nginx Kerberos/SPNEGO/etc. module Message-ID: All: If you're interested in this module, I am politely asking for some contributions towards it. I am going to probably ask one of the MIT Kerberos guys to do a quick scan over the code and tweak any Kerberos based stuff that could be optimized (he doesn't come cheap) and then possibly contract an nginx-capable developer to mature it from that side too. I've already put some funds in and I am willing to put a little bit more, but if this is something you want, please contact me, it would be great to have some of the best eyeballs looking at the code and mature it sooner rather than later. I also plan on releasing it and maybe sticking it on github or something, so that anyone can contribute and such, but before that I'd like to get it stable as it can be. :) Thanks, mike From nginx-forum at nginx.us Fri Jul 3 22:01:18 2009 From: nginx-forum at nginx.us (tmiskiew) Date: Fri, 3 Jul 2009 14:01:18 -0400 Subject: How to set Retry-After for 503? In-Reply-To: <1483953504@forum.nginx.org> References: <418c2d7dfb3f50f6248f078e1a3c2620.NginxMailingList@forum.nginx.org> <1483953504@forum.nginx.org> Message-ID: <9549f93de576c7b47c60306e52947cf5.NginxMailingList@forum.nginx.org> OK, I see. Can you please explain how to set Retry-After for the status code 503? Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3727,3734#msg-3734 From gmm at csdoc.com Fri Jul 3 23:49:41 2009 From: gmm at csdoc.com (Gena Makhomed) Date: Fri, 3 Jul 2009 22:49:41 +0300 Subject: How to set Retry-After for 503? In-Reply-To: <9549f93de576c7b47c60306e52947cf5.NginxMailingList@forum.nginx.org> References: <418c2d7dfb3f50f6248f078e1a3c2620.NginxMailingList@forum.nginx.org> <1483953504@forum.nginx.org> <9549f93de576c7b47c60306e52947cf5.NginxMailingList@forum.nginx.org> Message-ID: <708240094.20090703224941@csdoc.com> On Friday, July 3, 2009 at 21:01:18, tmiskiew wrote: t> Can you please explain how to set Retry-After for the status code 503? workaround via apache/mod_php: > cat 503.php "."\x0d\x0a". "503 Service Temporarily Unavailable"."\x0d\x0a". ""."\x0d\x0a". "

503 Service Temporarily Unavailable

"."\x0d\x0a". "

nginx"."\x0d\x0a". ""."\x0d\x0a". ""."\x0d\x0a" ); > cat headers HTTP/1.1 503 Service Temporarily Unavailable Server: nginx Date: Fri, 03 Jul 2009 19:48:11 GMT Content-Type: text/html; charset=UTF-8 Connection: keep-alive Retry-After: 600 Content-Length: 206 -- Best regards, Gena From gmm at csdoc.com Fri Jul 3 23:46:09 2009 From: gmm at csdoc.com (Gena Makhomed) Date: Fri, 3 Jul 2009 22:46:09 +0300 Subject: How to set Retry-After for 503? In-Reply-To: <9549f93de576c7b47c60306e52947cf5.NginxMailingList@forum.nginx.org> References: <418c2d7dfb3f50f6248f078e1a3c2620.NginxMailingList@forum.nginx.org> <1483953504@forum.nginx.org> <9549f93de576c7b47c60306e52947cf5.NginxMailingList@forum.nginx.org> Message-ID: <1498095573.20090703224609@csdoc.com> On Friday, July 3, 2009 at 21:01:18, tmiskiew wrote: t> Can you please explain how to set Retry-After for the status code 503? workaround via ngx_http_perl_module: > cat /etc/nginx/perl/erorr503.pm package error503; use warnings; use strict; use nginx; sub handler { my $r = shift; $r->status( 503 ); $r->header_out( 'Retry-After', 600 ); $r->send_http_header( 'text/html' ); $r->print( ""."\x0d\x0a". "503 Service Temporarily Unavailable"."\x0d\x0a". ""."\x0d\x0a". "

503 Service Temporarily Unavailable

"."\x0d\x0a". "

nginx"."\x0d\x0a". ""."\x0d\x0a". ""."\x0d\x0a" ); } 1; > cat /etc/nginx/conf/nginx.conf # ... http { server_tokens off; perl_modules /etc/nginx/perl; perl_require erorr503.pm; server { server_name error503; error_page 503 = @error503; location @error503 { perl error503::handler; } location /error503 { return 503; } } # ... > curl http://error503/error503 -D headers 503 Service Temporarily Unavailable

503 Service Temporarily Unavailable

nginx > cat headers HTTP/1.1 503 Service Temporarily Unavailable Server: nginx Date: Fri, 03 Jul 2009 19:30:47 GMT Content-Type: text/html; charset=utf-8 Transfer-Encoding: chunked Connection: keep-alive Retry-After: 600 -- Best regards, Gena From theglobeisnowdigital at gmail.com Sat Jul 4 01:22:01 2009 From: theglobeisnowdigital at gmail.com (Magdalena Glinkowski) Date: Fri, 3 Jul 2009 14:22:01 -0700 Subject: control flow of configuration and bootup : where do the $variables come from (fastcgi to nginx) Message-ID: <432c29390907031422v6eaa885ahc1ea858b3269aca7@mail.gmail.com> preface: I'm new to web servers and scripting. what is the control flow of the fastcgi to nginx boot sequence? both the fast_cgi configuration file and the nginx configuration file have $variable assignments, what is the original instantiater of these variables? context for question: i notice there is an "emerg" component that is unhappy with my configuration file, and the source code suggests it's with one of my variables with the ngx_http_variables code, the exact message was "unknown "query" variable" now I have resolved it (interestingly enough I have no "query" variable (maybe its the default no time for inspection now), however the first fast_cgi param was "$query-string" (also new to this "fastcgi")), with this context, this leads me to my open question.... both the fast_cgi configuration file and the nginx configuration file have $variable assignments, what is the original instantiater of these variables? Magdalena -------------- next part -------------- An HTML attachment was scrubbed... URL: From merlincorey at dc949.org Sat Jul 4 13:21:40 2009 From: merlincorey at dc949.org (merlin corey) Date: Sat, 4 Jul 2009 02:21:40 -0700 Subject: Wordpress MU Config In-Reply-To: <9f73ef880906292228p31778b88h637bafd50d3213cb@mail.gmail.com> References: <715653D554AB4FB688B344E6F2E6571A@shridesk> <4A442E6C.9010104@gmail.com> <9f73ef880906292228p31778b88h637bafd50d3213cb@mail.gmail.com> Message-ID: On Mon, Jun 29, 2009 at 10:28 PM, Hone Watson wrote: > Merlin, how does that config account for visitors with logged in > wordpress user cookies? > > With Supercache these users should not be directed to the supercached files: > Ah thanks, I knew there would be something. I have not tested this yet (far too late) but something like the following should be what we want (will verify later, again line breaks are my client): location / { if ($http_cookie ~* ?comment_author_|wordpress|wp-postpass_? ) { try_files $uri $uri/ /index.php?q=$uri; break; // Not sure if needed here } try_files $uri $uri/ /wp-content/cache/supercache/$http_host/$uriindex.html /index.php?q=$uri; } I'll also try to find a better way to do this but I think at his time you will definitely need an if somewhere to check cookie value, though I would like to move it out of main location if possible. > > On Fri, Jun 26, 2009 at 10:17 PM, merlin corey wrote: >> Also the line breaks are from my mail client truncating... ?I have it >> all one line, but I don't think it matters for nginx. >> >> On Fri, Jun 26, 2009 at 5:16 AM, merlin corey wrote: >>> I recently setup wordpress with supercache like so: >>> >>> location / { >>> ?try_files $uri >>> /wp-content/cache/supercache/$http_host/$uriindex.html $uri/ >>> /index.php?q=$uri; >>> } >>> >>> I believe this is an Igor-cringing-from-too-many-ifs-less >>> configuration, but I do not guarantee the lack of any cringes >>> whatsoever :P. >>> >>> On Thu, Jun 25, 2009 at 7:11 PM, Jim Ohlstein wrote: >>>> >>>> >>>> Shri @ DevLib.Org wrote: >>>>> >>>>> Hi there, >>>>> ?Does anyone have an optimized config for Wordpress MU with subdomain >>>>> support? >>>> >>>> I use the config at >>>> http://lumanau.web.id/2008/11/27/nginx-rewrite-rules-for-wordpressmu-with-wp-super-cache-plugin-enabled.html >>>> for MU with supercache. It seems to work. I may switch to Varnish or nginx >>>> fastcgi cache and then the rewrites are pretty trivial. If I recall >>>> correctly the rewrites are the same for subdomain or subdirectory >>>> configuration. It's a matter of a config setting in MU and a wildcard DNS >>>> entry. >>>> >>>> Note these are not "optimized" in that they contain lots of "if" statements >>>> and Igor would probably cringe... >>>>> >>>>> ?There are a lot of conflicting configs on the wordpress forums and I >>>>> could not find one on the Nginx Wiki - hence the request. >>>>> ?Thanks in advance! >>>>> ?Regards, >>>>> Shri >>>>> >>>> >>>> Jim >>>> >>>> >>>> >>> >> >> > > From slogster at gmail.com Sat Jul 4 16:46:54 2009 From: slogster at gmail.com (Momchil Ivanov) Date: Sat, 4 Jul 2009 14:46:54 +0200 Subject: can someone post their working nginx+moinmoin config? In-Reply-To: <20090703152604.GC31310@leitl.org> References: <20090703134238.GA31310@leitl.org> <20090703142021.GA3505@dotimes.com> <20090703152604.GC31310@leitl.org> Message-ID: <200907041446.58994.slogster@gmail.com> ?? Friday 03 July 2009 17:26:04 Eugen Leitl ??????: > On Fri, Jul 03, 2009 at 10:20:21PM +0800, Cherife Li wrote: > > Cliff Wells has the experience using Twisted server. It's said to be > > better. Although I haven't tried it out. :-) > > Thank you. Have you considered using fcgi for MoinMoin, and if yes, > did you reject it for a reason? I am using MoinMoin with mod_wsgi, here is my config: location /foo { wsgi_pass /www/foo/moin.wsgi; wsgi_pass_authorization off; wsgi_script_reloading on; wsgi_use_main_interpreter on; } -- PGP KeyID: 0x3118168B Keyserver: pgp.mit.edu Key fingerprint BB50 2983 0714 36DC D02E ?158A E03D 56DA 3118 168B -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 198 bytes Desc: This is a digitally signed message part. URL: From merlincorey at dc949.org Sat Jul 4 23:47:05 2009 From: merlincorey at dc949.org (merlin corey) Date: Sat, 4 Jul 2009 12:47:05 -0700 Subject: External health checking - OPTIONS returns 405 In-Reply-To: <10F8B021-CD78-450B-860F-0B0B8612E994@sapo.pt> References: <10F8B021-CD78-450B-860F-0B0B8612E994@sapo.pt> Message-ID: Hello, You can change HAProxy's health check request method to HEAD. This would be the easiest/best solution and still highly lightweight, in my opinion. If for some reason you cannot do that, I created an extremely simple module for nginx which just always replies with a "normal" OPTIONS reply. You can download this module from http://www.rumandcoke.org/nginx/noptions.tar.gz ; extract it (I put it in src/http/modules) and append --add-module=src/http/modules/noptions (or your path to the module) then rebuild NginX. This has not been tested or used since 0.6.3? and may require some small changes for 0.7/0.8 (but it might not). Thanks, Merlin On Fri, Jul 3, 2009 at 9:45 AM, Pedro Mata-Mouros Fonseca wrote: > Greetings, > > The following is an HAProxy option httpchk check request. It's purpose is to > check a server's health by means of issuing a very lightweight request to > it. It does this, by default, using OPTIONS (it's possible to configure it > to perform any type of request, but OPTIONS is aparently the most > lightweight). > > Nginx unfortunately returns a 405 to this. It's not really a big deal, but > is there anyway to "force" Nginx to return a 200 with an empty payload or as > small as possible? Would you still consider this a lightweight check to > perform every 2 or so seconds, on a few hundred reqs/sec website? > > 127.0.0.1 - - [03/Jul/2009:17:34:21 +0100] "OPTIONS / HTTP/1.0" 405 166 "-" > "-" > > Thanks for your time. > > Pedro. > > From nginx-forum at nginx.us Sun Jul 5 13:40:15 2009 From: nginx-forum at nginx.us (batrick) Date: Sun, 5 Jul 2009 05:40:15 -0400 Subject: Variables in access_log In-Reply-To: <911f7c981591b165a63cb62d3117b09a.NginxMailingList@forum.nginx.org> References: <911f7c981591b165a63cb62d3117b09a.NginxMailingList@forum.nginx.org> Message-ID: I have run into a similar problem. I wanted to set up a number of virtual hosts that each share the same configuration except, of course, for the domain name and path. So my plan was to set some environment variables, and then include the template configuration file. Unfortunately, the plan faltered somewhat because error_log and server_name do not seem to support variables. Does anybody know of a documentation which directives support variables and which do not? Kind regards, batrick Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3722,3763#msg-3763 From andrew at andrewloe.com Sun Jul 5 21:14:34 2009 From: andrew at andrewloe.com (W. Andrew Loe III) Date: Sun, 5 Jul 2009 10:14:34 -0700 Subject: Advanced Maintenance In-Reply-To: <20090702152458.GS5360@mdounin.ru> References: <99e8cb160906302309q37c5b209o17fd68b1f52bbdb@mail.gmail.com> <20090701093047.GC5360@mdounin.ru> <4A4B935D.3030803@gmail.com> <013d01c9fa73$9c781bf0$d56853d0$@com> <20090701231921.GL5360@mdounin.ru> <024b01c9fac2$0e91b7c0$2bb52740$@com> <02d601c9fb13$fd96aaf0$f8c400d0$@com> <20090702152458.GS5360@mdounin.ru> Message-ID: <9007bd350907051014s716c529cxe14a75dfc104c3de@mail.gmail.com> I wrote a little article on doing this: http://onehub.com/past/2009/3/6/rails_maintenance_pages_done_right/ On Thu, Jul 2, 2009 at 8:24 AM, Maxim Dounin wrote: > Hello! > > On Thu, Jul 02, 2009 at 07:52:42AM -0500, AMP Admin wrote: > >> Ohhh... i was looking for a not operator like '!'. ?thanks, I'll give that a try! > > It's much better to just configure normal more specific locations instead > of using regexps, e.g. > > location / { > ? ?# maintanance checks here > } > > location /maintanance.php { > ? ?# no checks here > } > > location /privacypolicy.php { > ? ?# and no checks here > } > > location /images/ { > ? ?# and no checks here > } > > location /promo/ { > ? ?# and no checks here > } > > Each request will be processed in most-specific location, and > hence there would be no maintanance checks for privacypolicy.php, > images and promo (and maintanance.php itself). > > Maxim Dounin > > >> >> -----Original Message----- >> From: owner-nginx at sysoev.ru [mailto:owner-nginx at sysoev.ru] On Behalf Of Edho P Arief >> Sent: Wednesday, July 01, 2009 10:38 PM >> To: nginx at sysoev.ru >> Subject: Re: Advanced Maintenance >> >> On Thu, Jul 2, 2009 at 10:06 AM, AMP Admin wrote: >> > I don't understand. ?How do we make the "no maintenance checks"? ?sorry, I'm >> > still learning. >> > >> > If we want everything but privacypolicy.php, images directory and promo >> > directory to go to maintenance how would that look? >> > >> > It seems that this puts the entire site in maintenance mode: >> > Location / { >> > ? ? ? ?if (-f $document_root/maintenance.php) { >> > ? ? ? ? ? ? ? ?rewrite ?^(.*)$ ?/maintenance.php last; >> > ? ? ? ? ? ? ? ?break; >> > ? ? ? ?} >> > } >> > >> > I'm not sure how to do the no maintenance part. >> > >> >> location ~ ^/(?!privacypolicy.php|images|promo) { >> ? try_files /maintenance.php $uri @fallback; >> } >> >> -- >> O< ascii ribbon campaign - stop html mail - www.asciiribbon.org >> >> > > From tlott at gamesnet.de Mon Jul 6 07:37:27 2009 From: tlott at gamesnet.de (Tobias Lott) Date: Mon, 6 Jul 2009 05:37:27 +0200 Subject: Did I got try_files wrong? and bad gateway 502 Message-ID: <20090706053727.43700b5d@sub.han.vpn.gamesnet.de> Hey Folks I got a problem understanding try_files I guess. Got this as configuration : location = / { rewrite ^(.*)$ /shop/ redirect; } location / { proxy_pass http://backend:10060; proxy_set_header X-Real-IP $remote_addr; proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for; proxy_set_header Host $http_host; proxy_redirect off; proxy_connect_timeout 90; proxy_send_timeout 90; proxy_read_timeout 120; } Problem is that I cant "catch" all 404 and rewrite it to /shop/index.php but with "if" (which I shouldn't use cause of possible performance issues, Igor mentioned I think.) Doesn't work: try_files $uri $uri/ /shop/index.php; error_page 404 /shop/index.php; Works: if (!-f $request_filename) { rewrite ^(.*)$ /shop/index.php redirect; } Additionally I got another issue, which is that customers getting 502s pretty damn randomly (using nginx-0.7.61 as frontend and apache 2.2.11 as backend for the time being, which gonna gets migrated soon). Was trying to tweak a lil with proxy_*_timeout and keepalive_timeout but since its not reproducible I can't be sure its fixed, are there some "good to go" values or another option to check? Thanks so much Igor for that godly piece of software and all involved. Greetings -- Tobias Lott From nbubingo at gmail.com Mon Jul 6 07:38:04 2009 From: nbubingo at gmail.com (Weibin Yao) Date: Mon, 06 Jul 2009 11:38:04 +0800 Subject: 403:wiki.codemongers.com Message-ID: <4A51719C.3000904@gmail.com> 403 Forbidden ------------------------------------------------------------------------ nginx/0.7.51 -- Weibin Yao From cherife at dotimes.com Mon Jul 6 08:20:34 2009 From: cherife at dotimes.com (Cherife Li) Date: Mon, 6 Jul 2009 12:20:34 +0800 Subject: 403:wiki.codemongers.com In-Reply-To: <4A51719C.3000904@gmail.com> References: <4A51719C.3000904@gmail.com> Message-ID: <20090706042034.GA1647@dotimes.com> On Mon, 06 Jul 2009 at 11:38:04 +0800, Weibin Yao wrote: > > 403 Forbidden > That's the old wiki. Check the new one at . -- Rgds, Cherife. -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 197 bytes Desc: not available URL: From mike503 at gmail.com Mon Jul 6 08:39:44 2009 From: mike503 at gmail.com (Michael Shadle) Date: Sun, 5 Jul 2009 21:39:44 -0700 Subject: Did I got try_files wrong? and bad gateway 502 In-Reply-To: <20090706053727.43700b5d@sub.han.vpn.gamesnet.de> References: <20090706053727.43700b5d@sub.han.vpn.gamesnet.de> Message-ID: On Sun, Jul 5, 2009 at 8:37 PM, Tobias Lott wrote: > Additionally I got another issue, which is that customers getting 502s > pretty damn randomly (using nginx-0.7.61 as frontend and apache 2.2.11 > as backend for the time being, which gonna gets migrated soon). Was > trying to tweak a lil with proxy_*_timeout and keepalive_timeout but > since its not reproducible I can't be sure its fixed, are there some > "good to go" values or another option to check? Is it nginx giving the 502 or apache? From jg at internetx.de Mon Jul 6 12:28:10 2009 From: jg at internetx.de (Juergen Gotteswinter) Date: Mon, 06 Jul 2009 10:28:10 +0200 Subject: Nginx Punycode Support Message-ID: <4A51B59A.9030208@internetx.de> Hello does nginx support punycode domains in the server_name directives? thanks From tlott at gamesnet.de Mon Jul 6 14:03:41 2009 From: tlott at gamesnet.de (Tobias Lott) Date: Mon, 6 Jul 2009 12:03:41 +0200 Subject: Did I got try_files wrong? and bad gateway 502 In-Reply-To: References: <20090706053727.43700b5d@sub.han.vpn.gamesnet.de> Message-ID: <20090706120341.5ab5e64f@sub.han.vpn.gamesnet.de> On Sun, 5 Jul 2009 21:39:44 -0700 Michael Shadle wrote: > On Sun, Jul 5, 2009 at 8:37 PM, Tobias Lott wrote: > > > Additionally I got another issue, which is that customers getting > > 502s pretty damn randomly (using nginx-0.7.61 as frontend and > > apache 2.2.11 as backend for the time being, which gonna gets > > migrated soon). Was trying to tweak a lil with proxy_*_timeout and > > keepalive_timeout but since its not reproducible I can't be sure > > its fixed, are there some "good to go" values or another option to > > check? > > Is it nginx giving the 502 or apache? > nginx is giving the 502 -- Tobias Lott From juangiordana at gmail.com Mon Jul 6 18:50:19 2009 From: juangiordana at gmail.com (Juan Fco. Giordana) Date: Mon, 06 Jul 2009 11:50:19 -0300 Subject: 403:wiki.codemongers.com In-Reply-To: <20090706042034.GA1647@dotimes.com> References: <4A51719C.3000904@gmail.com> <20090706042034.GA1647@dotimes.com> Message-ID: <4A520F2B.1020405@gmail.com> A 301 redirect to http://wiki.nginx.org/ should be more appropriate then :) Cherife Li wrote: > On Mon, 06 Jul 2009 at 11:38:04 +0800, Weibin Yao wrote: > >> 403 Forbidden >> > That's the old wiki. > Check the new one at . From lists at ruby-forum.com Mon Jul 6 19:23:02 2009 From: lists at ruby-forum.com (Mi Mi) Date: Mon, 6 Jul 2009 17:23:02 +0200 Subject: Can't upload files > 10Mb In-Reply-To: <20080515173919.GA27880@rambler-co.ru> References: <1210869319.4449.12.camel@portableevil.develix.com> <156c5bc40805151001u75e8b4c7k12994b7e03935aca@mail.gmail.com> <1210872776.4449.17.camel@portableevil.develix.com> <20080515173919.GA27880@rambler-co.ru> Message-ID: <00a1b89ebf69500764d1b63939c4c7ec@ruby-forum.com> Igor Sysoev wrote: > On Thu, May 15, 2008 at 10:32:56AM -0700, Cliff Wells wrote: > >> >> On Thu, 2008-05-15 at 13:01 -0400, Tieg Zaharia wrote: >> > Hey Cliff-- I had the same problem a few weeks ago and then realized >> > that the "client_max_body_size" parameter wasn't in the right place. >> > Do you have it under a location block? >> >> It's in the "http" section. I'll try adding it to the specific location >> block. > > Is this error still in error_log ? > What is your backend ? It may its own limit. I have the same problem and I don't have any explicit "max" setting in any configuration file. I'm not sure what you mean by "backend" -- but I cast "http". [root at myhost /usr/local/etc/nginx]# grep -ri max . [root at myhost /usr/local/etc/nginx]# grep -ri upload . [root at myhost /usr/local/etc/nginx]# Although I read here https://cloud-support.engineyard.com/discussions/problems/228-nginxpassenger-issue-file-uploads-not-working-client-intended-to-send-too-large-body that "upload_max_file_size 50m; client_max_body_size 50m;" fix the problem, my nginx 0.7.61 refuses to accept the former in http { }. Also, I read from http://wiki.nginx.org/NginxHttpCoreModule#client_max_body_size that the default size is 1m. Is it reasonable to specify a maximum by default, especially *this* strict? For giving some context: I stumbled into the problem when people were reporting that the majority, but not all ("indeterministically", as they said) of the Document Uploads in an OpenGoo installation did not work. -- Posted via http://www.ruby-forum.com/. From cliff at develix.com Mon Jul 6 21:10:50 2009 From: cliff at develix.com (Cliff Wells) Date: Mon, 06 Jul 2009 10:10:50 -0700 Subject: Can't upload files > 10Mb In-Reply-To: <00a1b89ebf69500764d1b63939c4c7ec@ruby-forum.com> References: <1210869319.4449.12.camel@portableevil.develix.com> <156c5bc40805151001u75e8b4c7k12994b7e03935aca@mail.gmail.com> <1210872776.4449.17.camel@portableevil.develix.com> <20080515173919.GA27880@rambler-co.ru> <00a1b89ebf69500764d1b63939c4c7ec@ruby-forum.com> Message-ID: <1246900250.4626.1.camel@portable-evil> On Mon, 2009-07-06 at 17:23 +0200, Mi Mi wrote: > Igor Sysoev wrote: > > On Thu, May 15, 2008 at 10:32:56AM -0700, Cliff Wells wrote: > > > >> > >> On Thu, 2008-05-15 at 13:01 -0400, Tieg Zaharia wrote: > >> > Hey Cliff-- I had the same problem a few weeks ago and then realized > >> > that the "client_max_body_size" parameter wasn't in the right place. > >> > Do you have it under a location block? > >> > >> It's in the "http" section. I'll try adding it to the specific location > >> block. > > > > Is this error still in error_log ? > > What is your backend ? It may its own limit. > > I have the same problem and I don't have any explicit "max" setting in > any configuration file. I'm not sure what you mean by "backend" -- but I > cast "http". There's a default value of 1MB for client_max_body_size, so if you don't have any explicit setting then that's your limit: http://wiki.nginx.org/NginxHttpCoreModule#client_max_body_size Cliff -- http://www.google.com/search?q=vonage+sucks From mike503 at gmail.com Tue Jul 7 00:29:58 2009 From: mike503 at gmail.com (Michael Shadle) Date: Mon, 6 Jul 2009 13:29:58 -0700 Subject: Did I got try_files wrong? and bad gateway 502 In-Reply-To: <20090706120341.5ab5e64f@sub.han.vpn.gamesnet.de> References: <20090706053727.43700b5d@sub.han.vpn.gamesnet.de> <20090706120341.5ab5e64f@sub.han.vpn.gamesnet.de> Message-ID: Check the nginx error log for what the message is when you hit those 502's. Perhaps apache is too busy to service the request or is sending malformed resposne etc? On Mon, Jul 6, 2009 at 3:03 AM, Tobias Lott wrote: > > > On Sun, 5 Jul 2009 21:39:44 -0700 > Michael Shadle wrote: > >> On Sun, Jul 5, 2009 at 8:37 PM, Tobias Lott wrote: >> >> > Additionally I got another issue, which is that customers getting >> > 502s pretty damn randomly (using nginx-0.7.61 as frontend and >> > apache 2.2.11 as backend for the time being, which gonna gets >> > migrated soon). Was trying to tweak a lil with proxy_*_timeout and >> > keepalive_timeout but since its not reproducible I can't be sure >> > its fixed, are there some "good to go" values or another option to >> > check? >> >> Is it nginx giving the 502 or apache? >> > > nginx is giving the 502 > > -- > Tobias Lott > > From tlott at gamesnet.de Tue Jul 7 01:52:02 2009 From: tlott at gamesnet.de (Tobias Lott) Date: Mon, 6 Jul 2009 23:52:02 +0200 Subject: Did I got try_files wrong? and bad gateway 502 In-Reply-To: References: <20090706053727.43700b5d@sub.han.vpn.gamesnet.de> <20090706120341.5ab5e64f@sub.han.vpn.gamesnet.de> Message-ID: <20090706235202.7b12b96d@sub.han.vpn.gamesnet.de> On Mon, 6 Jul 2009 13:29:58 -0700 Michael Shadle wrote: > Check the nginx error log for what the message is when you hit those > 502's. > > Perhaps apache is too busy to service the request or is sending > malformed resposne etc? > > On Mon, Jul 6, 2009 at 3:03 AM, Tobias Lott wrote: > > > > > > On Sun, 5 Jul 2009 21:39:44 -0700 > > Michael Shadle wrote: > > > >> On Sun, Jul 5, 2009 at 8:37 PM, Tobias Lott > >> wrote: > >> > >> > Additionally I got another issue, which is that customers getting > >> > 502s pretty damn randomly (using nginx-0.7.61 as frontend and > >> > apache 2.2.11 as backend for the time being, which gonna gets > >> > migrated soon). Was trying to tweak a lil with proxy_*_timeout > >> > and keepalive_timeout but since its not reproducible I can't be > >> > sure its fixed, are there some "good to go" values or another > >> > option to check? > >> > >> Is it nginx giving the 502 or apache? > >> > > > > nginx is giving the 502 > > > > -- > > Tobias Lott > > > > > The only error log entry thats consistent with the 502 is the following, but its not the same file, one type of file, same client: 2009/07/06 22:48:20 [crit] 86851#0: *183664 connect() to XX.XX.XX.XX:XX failed (1: Operation not permitted) while connecting to upstream, client: XX.XX.XX.XX, server: hostname.tld, request: "GET /some/random/picture.jpg HTTP/1.1", upstream: "http://XX.XX.XX.XX:XX/some/random/picture.jpg", host: "hostname.tld", referrer: "http://hostname.tld/some/" There is nothing in Apache error/access log at that time. -- Tobias Lott From pedro.matamouros at sapo.pt Tue Jul 7 03:58:18 2009 From: pedro.matamouros at sapo.pt (Pedro Mata-Mouros Fonseca) Date: Tue, 7 Jul 2009 00:58:18 +0100 Subject: External health checking - OPTIONS returns 405 In-Reply-To: References: <10F8B021-CD78-450B-860F-0B0B8612E994@sapo.pt> Message-ID: <9BE95041-C67A-47B6-B7AF-B7F3DDB61956@sapo.pt> Many thanks for the sugestions Merlin, going to check out the HEAD method one. Pedro. On Jul 4, 2009, at 8:47 PM, merlin corey wrote: > Hello, > > You can change HAProxy's health check request method to HEAD. This > would be the easiest/best solution and still highly lightweight, in my > opinion. > > If for some reason you cannot do that, I created an extremely simple > module for nginx which just always replies with a "normal" OPTIONS > reply. You can download this module from > http://www.rumandcoke.org/nginx/noptions.tar.gz ; extract it (I put it > in src/http/modules) and append --add-module=src/http/modules/noptions > (or your path to the module) then rebuild NginX. This has not been > tested or used since 0.6.3? and may require some small changes for > 0.7/0.8 (but it might not). > > Thanks, > Merlin > > On Fri, Jul 3, 2009 at 9:45 AM, Pedro Mata-Mouros > Fonseca wrote: >> Greetings, >> >> The following is an HAProxy option httpchk check request. It's >> purpose is to >> check a server's health by means of issuing a very lightweight >> request to >> it. It does this, by default, using OPTIONS (it's possible to >> configure it >> to perform any type of request, but OPTIONS is aparently the most >> lightweight). >> >> Nginx unfortunately returns a 405 to this. It's not really a big >> deal, but >> is there anyway to "force" Nginx to return a 200 with an empty >> payload or as >> small as possible? Would you still consider this a lightweight >> check to >> perform every 2 or so seconds, on a few hundred reqs/sec website? >> >> 127.0.0.1 - - [03/Jul/2009:17:34:21 +0100] "OPTIONS / HTTP/1.0" 405 >> 166 "-" >> "-" >> >> Thanks for your time. >> >> Pedro. >> >> > > From merlincorey at dc949.org Tue Jul 7 04:41:40 2009 From: merlincorey at dc949.org (merlin corey) Date: Mon, 6 Jul 2009 17:41:40 -0700 Subject: Variables in access_log In-Reply-To: References: <911f7c981591b165a63cb62d3117b09a.NginxMailingList@forum.nginx.org> Message-ID: I believe server_name now supports some variables but maybe this is just in development branch. The official documentation is in Russian and located at http://sysoev.ru/nginx/docs/ . There is also a wiki at http://wiki.nginx.org with English and Chinese translations of some of the documentation, but it is not always up to date or complete. Google translate can do a readable job in many cases, and the configuration language is always the same in any language, plus there is the common language C that nginx is written in. These are the sources you should look through. Completely separately regarding configurations, I am under the understanding that nginx performs better with fully specified server names. It should be easy enough to put common configuration into files which you include in each server { } block as necessary. This can easily be generated by a script so you don't have to do it by hand if you have a lot of domains. The benefits are you get the common configuration, separate error logging, separate access logging, and flexibility to make individual host changes while efficiently selecting proper server blocks. -- Merlin On 7/5/09, batrick wrote: > I have run into a similar problem. I wanted to set up a number of virtual > hosts that each share the same configuration except, of course, for the > domain name and path. So my plan was to set some environment variables, and > then include the template configuration file. > > Unfortunately, the plan faltered somewhat because error_log and server_name > do not seem to support variables. > > Does anybody know of a documentation which directives support variables and > which do not? > > Kind regards, > > batrick > > Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3722,3763#msg-3763 > > > From merlincorey at dc949.org Tue Jul 7 04:43:43 2009 From: merlincorey at dc949.org (merlin corey) Date: Mon, 6 Jul 2009 17:43:43 -0700 Subject: Advanced Maintenance In-Reply-To: <9007bd350907051014s716c529cxe14a75dfc104c3de@mail.gmail.com> References: <99e8cb160906302309q37c5b209o17fd68b1f52bbdb@mail.gmail.com> <20090701093047.GC5360@mdounin.ru> <4A4B935D.3030803@gmail.com> <013d01c9fa73$9c781bf0$d56853d0$@com> <20090701231921.GL5360@mdounin.ru> <024b01c9fac2$0e91b7c0$2bb52740$@com> <02d601c9fb13$fd96aaf0$f8c400d0$@com> <20090702152458.GS5360@mdounin.ru> <9007bd350907051014s716c529cxe14a75dfc104c3de@mail.gmail.com> Message-ID: You should write a new article with an updated try_files method with 0.7.x being stable now! ;) On 7/5/09, W. Andrew Loe III wrote: > I wrote a little article on doing this: > > http://onehub.com/past/2009/3/6/rails_maintenance_pages_done_right/ > > On Thu, Jul 2, 2009 at 8:24 AM, Maxim Dounin wrote: >> Hello! >> >> On Thu, Jul 02, 2009 at 07:52:42AM -0500, AMP Admin wrote: >> >>> Ohhh... i was looking for a not operator like '!'. ?thanks, I'll give >>> that a try! >> >> It's much better to just configure normal more specific locations instead >> of using regexps, e.g. >> >> location / { >> ? ?# maintanance checks here >> } >> >> location /maintanance.php { >> ? ?# no checks here >> } >> >> location /privacypolicy.php { >> ? ?# and no checks here >> } >> >> location /images/ { >> ? ?# and no checks here >> } >> >> location /promo/ { >> ? ?# and no checks here >> } >> >> Each request will be processed in most-specific location, and >> hence there would be no maintanance checks for privacypolicy.php, >> images and promo (and maintanance.php itself). >> >> Maxim Dounin >> >> >>> >>> -----Original Message----- >>> From: owner-nginx at sysoev.ru [mailto:owner-nginx at sysoev.ru] On Behalf Of >>> Edho P Arief >>> Sent: Wednesday, July 01, 2009 10:38 PM >>> To: nginx at sysoev.ru >>> Subject: Re: Advanced Maintenance >>> >>> On Thu, Jul 2, 2009 at 10:06 AM, AMP Admin wrote: >>> > I don't understand. ?How do we make the "no maintenance checks"? >>> > ?sorry, I'm >>> > still learning. >>> > >>> > If we want everything but privacypolicy.php, images directory and promo >>> > directory to go to maintenance how would that look? >>> > >>> > It seems that this puts the entire site in maintenance mode: >>> > Location / { >>> > ? ? ? ?if (-f $document_root/maintenance.php) { >>> > ? ? ? ? ? ? ? ?rewrite ?^(.*)$ ?/maintenance.php last; >>> > ? ? ? ? ? ? ? ?break; >>> > ? ? ? ?} >>> > } >>> > >>> > I'm not sure how to do the no maintenance part. >>> > >>> >>> location ~ ^/(?!privacypolicy.php|images|promo) { >>> ? try_files /maintenance.php $uri @fallback; >>> } >>> >>> -- >>> O< ascii ribbon campaign - stop html mail - www.asciiribbon.org >>> >>> >> >> > > From merlincorey at dc949.org Tue Jul 7 04:50:00 2009 From: merlincorey at dc949.org (merlin corey) Date: Mon, 6 Jul 2009 17:50:00 -0700 Subject: Nginx Punycode Support In-Reply-To: <4A51B59A.9030208@internetx.de> References: <4A51B59A.9030208@internetx.de> Message-ID: Hello, NginX supports regular expressions, unix globs, and such within server_name directive (see http://sysoev.ru/nginx/docs/http/ngx_http_core_module.html#server_name and http://wiki.nginx.org/NginxHttpCoreModule#server_name )... Whether or not you can use that for your unicode encoding, I am not sure :). Thanks, Merlin On 7/6/09, Juergen Gotteswinter wrote: > Hello > > does nginx support punycode domains in the server_name directives? > > thanks > > > > > From merlincorey at dc949.org Tue Jul 7 04:53:25 2009 From: merlincorey at dc949.org (merlin corey) Date: Mon, 6 Jul 2009 17:53:25 -0700 Subject: Did I got try_files wrong? and bad gateway 502 In-Reply-To: <20090706235202.7b12b96d@sub.han.vpn.gamesnet.de> References: <20090706053727.43700b5d@sub.han.vpn.gamesnet.de> <20090706120341.5ab5e64f@sub.han.vpn.gamesnet.de> <20090706235202.7b12b96d@sub.han.vpn.gamesnet.de> Message-ID: Why not server some random picture from NginX? Surely that will relieve the 502s. On 7/6/09, Tobias Lott wrote: > > > On Mon, 6 Jul 2009 13:29:58 -0700 > Michael Shadle wrote: > >> Check the nginx error log for what the message is when you hit those >> 502's. >> >> Perhaps apache is too busy to service the request or is sending >> malformed resposne etc? >> >> On Mon, Jul 6, 2009 at 3:03 AM, Tobias Lott wrote: >> > >> > >> > On Sun, 5 Jul 2009 21:39:44 -0700 >> > Michael Shadle wrote: >> > >> >> On Sun, Jul 5, 2009 at 8:37 PM, Tobias Lott >> >> wrote: >> >> >> >> > Additionally I got another issue, which is that customers getting >> >> > 502s pretty damn randomly (using nginx-0.7.61 as frontend and >> >> > apache 2.2.11 as backend for the time being, which gonna gets >> >> > migrated soon). Was trying to tweak a lil with proxy_*_timeout >> >> > and keepalive_timeout but since its not reproducible I can't be >> >> > sure its fixed, are there some "good to go" values or another >> >> > option to check? >> >> >> >> Is it nginx giving the 502 or apache? >> >> >> > >> > nginx is giving the 502 >> > >> > -- >> > Tobias Lott >> > >> > >> > > The only error log entry thats consistent with the 502 is the > following, but its not the same file, one type of file, same client: > > 2009/07/06 22:48:20 [crit] 86851#0: *183664 connect() to > XX.XX.XX.XX:XX failed (1: Operation not permitted) while connecting > to upstream, client: XX.XX.XX.XX, server: hostname.tld, request: > "GET /some/random/picture.jpg HTTP/1.1", upstream: > "http://XX.XX.XX.XX:XX/some/random/picture.jpg", > host: "hostname.tld", referrer: "http://hostname.tld/some/" > > > There is nothing in Apache error/access log at that time. > > > -- > Tobias Lott > > From mdounin at mdounin.ru Tue Jul 7 06:06:10 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Tue, 7 Jul 2009 06:06:10 +0400 Subject: Did I got try_files wrong? and bad gateway 502 In-Reply-To: <20090706235202.7b12b96d@sub.han.vpn.gamesnet.de> References: <20090706053727.43700b5d@sub.han.vpn.gamesnet.de> <20090706120341.5ab5e64f@sub.han.vpn.gamesnet.de> <20090706235202.7b12b96d@sub.han.vpn.gamesnet.de> Message-ID: <20090707020609.GS5360@mdounin.ru> Hello! On Mon, Jul 06, 2009 at 11:52:02PM +0200, Tobias Lott wrote: > > > On Mon, 6 Jul 2009 13:29:58 -0700 > Michael Shadle wrote: > > > Check the nginx error log for what the message is when you hit those > > 502's. > > > > Perhaps apache is too busy to service the request or is sending > > malformed resposne etc? > > > > On Mon, Jul 6, 2009 at 3:03 AM, Tobias Lott wrote: > > > > > > > > > On Sun, 5 Jul 2009 21:39:44 -0700 > > > Michael Shadle wrote: > > > > > >> On Sun, Jul 5, 2009 at 8:37 PM, Tobias Lott > > >> wrote: > > >> > > >> > Additionally I got another issue, which is that customers getting > > >> > 502s pretty damn randomly (using nginx-0.7.61 as frontend and > > >> > apache 2.2.11 as backend for the time being, which gonna gets > > >> > migrated soon). Was trying to tweak a lil with proxy_*_timeout > > >> > and keepalive_timeout but since its not reproducible I can't be > > >> > sure its fixed, are there some "good to go" values or another > > >> > option to check? > > >> > > >> Is it nginx giving the 502 or apache? > > >> > > > > > > nginx is giving the 502 > > > > > > -- > > > Tobias Lott > > > > > > > > > > The only error log entry thats consistent with the 502 is the > following, but its not the same file, one type of file, same client: > > 2009/07/06 22:48:20 [crit] 86851#0: *183664 connect() to > XX.XX.XX.XX:XX failed (1: Operation not permitted) while connecting > to upstream, client: XX.XX.XX.XX, server: hostname.tld, request: > "GET /some/random/picture.jpg HTTP/1.1", upstream: > "http://XX.XX.XX.XX:XX/some/random/picture.jpg", > host: "hostname.tld", referrer: "http://hostname.tld/some/" This is EPERM, and most typical reasons include statefull firewall between nginx frontend and backend or timewait reuse switched off. If it's statefull firewall - remove it or convert to non-statefull one. If it's timewait issue - switch on timewait reuse/recycle (see your OS tuning guides to find out how). > There is nothing in Apache error/access log at that time. Since nginx wasn't allowed to connect() - it's somewhat expected. Maxim Dounin From lists at ruby-forum.com Tue Jul 7 09:33:30 2009 From: lists at ruby-forum.com (Jauder Ho) Date: Tue, 7 Jul 2009 07:33:30 +0200 Subject: try_files and static web site Message-ID: <68dea8f82aef5a4f9d2f1d4e02a49f5b@ruby-forum.com> I have a site that is completely static and I figured I would use try_files however this results in HTTP 500 errors for directories i.e. trying to get to http://foo.com/bar/ results in a 500 error. What's wrong with this config? Thanks. --Jauder server { # port to listen on. Can also be set to an IP:PORT listen 80; # Set the charset charset utf-8; # sets the domain[s] that this vhost server requests for server_name www.foo.com; # doc root root /var/www/foo.com; # vhost specific access log access_log /var/log/nginx_access.log main; # Catchall for everything else location / { root /var/www/foo.com; access_log off; index index.html index.shtml; expires 1d; try_files $uri $uri/; #if (-f $request_filename) { # break; #} } error_page 500 502 503 504 /500.html; error_page 404 /404.html; } -- Posted via http://www.ruby-forum.com/. From mike503 at gmail.com Tue Jul 7 09:47:17 2009 From: mike503 at gmail.com (Michael Shadle) Date: Mon, 6 Jul 2009 22:47:17 -0700 Subject: try_files and static web site In-Reply-To: <68dea8f82aef5a4f9d2f1d4e02a49f5b@ruby-forum.com> References: <68dea8f82aef5a4f9d2f1d4e02a49f5b@ruby-forum.com> Message-ID: <53F14C2F-996A-4EAB-B0CF-AC209C2B01E0@gmail.com> I see no reason to be using try_files at all in your config. Sent from my iPhone On Jul 6, 2009, at 10:33 PM, Jauder Ho wrote: > I have a site that is completely static and I figured I would use > try_files however this results in HTTP 500 errors for directories i.e. > trying to get to http://foo.com/bar/ results in a 500 error. What's > wrong with this config? Thanks. > > --Jauder > > server { > # port to listen on. Can also be set to an IP:PORT > listen 80; > > # Set the charset > charset utf-8; > > # sets the domain[s] that this vhost server requests for > server_name www.foo.com; > > # doc root > root /var/www/foo.com; > > # vhost specific access log > access_log /var/log/nginx_access.log main; > > > # Catchall for everything else > location / { > root /var/www/foo.com; > access_log off; > > index index.html index.shtml; > expires 1d; > > try_files $uri $uri/; > > #if (-f $request_filename) { > # break; > #} > } > > > > error_page 500 502 503 504 /500.html; > error_page 404 /404.html; > > } > -- > Posted via http://www.ruby-forum.com/. > From nginx-forum at nginx.us Tue Jul 7 13:44:56 2009 From: nginx-forum at nginx.us (miradev) Date: Tue, 7 Jul 2009 05:44:56 -0400 Subject: Variables in access_log In-Reply-To: <4A4E1918.9050909@forum.nginx.org> References: <911f7c981591b165a63cb62d3117b09a.NginxMailingList@forum.nginx.org> <4A4E1918.9050909@forum.nginx.org> Message-ID: <52305158abd4031fc068f881cce01669.NginxMailingList@forum.nginx.org> Brent Bloxam Wrote: > > Have I misunderstood? > > Yes, you assumed error_log was the same as > access_log. They are not. > access_log is part of the Log module while > error_log is part of the main > module. error_log (at least in 0.7.59) does not > support variables in the > path. You are right, I missed that completely. Strangely this still fits the requirements, so it is not an issue for me. Thank you. Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3722,3806#msg-3806 From nginx-forum at nginx.us Tue Jul 7 13:53:02 2009 From: nginx-forum at nginx.us (miradev) Date: Tue, 7 Jul 2009 05:53:02 -0400 Subject: Variables in access_log In-Reply-To: References: <911f7c981591b165a63cb62d3117b09a.NginxMailingList@forum.nginx.org> Message-ID: <54705fcdd9412b89afaabed6dbb4541a.NginxMailingList@forum.nginx.org> batrick Wrote: ------------------------------------------------------- > I have run into a similar problem. I wanted to set > up a number of virtual hosts that each share the > same configuration except, of course, for the > domain name and path. So my plan was to set some > environment variables, and then include the > template configuration file. My particular set up is not for 'live' sites (in the typical manner), and I would support Merlin's suggestion to build config files for each one. I do infact do exactly that. A simple template with some variables replaced, and you are away. You never know when you need to add just that one custom setting for that client. For the dev sites, I would just build a completely new site if something specific was wanted, but to unravel a live site from such a fixed structure might be troublesome. > Unfortunately, the plan faltered somewhat because > error_log and server_name do not seem to support > variables. > > Does anybody know of a documentation which > directives support variables and which do not? I'm guessing with the pace at which nginx is moving, that if you want such a list, you will have to head to the wiki and compile it yourself. However, server_name does take wild cards, and you can do regular expression matching on $host. As I have done in my original post. Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3722,3807#msg-3807 From nginx-forum at nginx.us Tue Jul 7 13:59:31 2009 From: nginx-forum at nginx.us (davidc) Date: Tue, 7 Jul 2009 05:59:31 -0400 Subject: Logging embedded Perl return value In-Reply-To: <20090410103340.GR21297@forum.nginx.org> References: <20090409051639.GA87521@forum.nginx.org> <395c82ca4719d4e018910bbe6608ce1e.NginxMailingList@forum.nginx.org> <20090410103340.GR21297@forum.nginx.org> Message-ID: <227390a136e3ae4911bdf2352190d802.NginxMailingList@forum.nginx.org> Hi Igor, Thanks for the patch! I just realized it's now in the stable 0.7.x release. As a test, I tried adding a '207 response code by modifying the files below and return 'HTTP_CUSTOM_RESPONSE' from the perl handler. It didn't work so perhaps you could point me to where I'm going wrong? diff -r orig/nginx-0.7.61/src/http/modules/perl/nginx.pm nginx-0.7.61/src/http/modules/perl/nginx.pm 18a19 > HTTP_CUSTOM_RESPONSE 63a65 > use constant HTTP_CUSTOM_RESPONSE => 207; diff -r orig/nginx-0.7.61/src/http/ngx_http_header_filter_module.c nginx-0.7.61/src/http/ngx_http_header_filter_module.c 60a61 > ngx_string("207 Custom Response"), 64c65 < #define NGX_HTTP_LAST_LEVEL_200 207 --- > #define NGX_HTTP_LAST_LEVEL_200 208 diff -r orig/nginx-0.7.61/src/http/ngx_http_request.h nginx-0.7.61/src/http/ngx_http_request.h 68a69 > #define NGX_HTTP_CUSTOM_RESPONSE 207 Posted at Nginx Forum: http://forum.nginx.org/read.php?2,933,3808#msg-3808 From jg at internetx.de Tue Jul 7 15:36:27 2009 From: jg at internetx.de (Juergen Gotteswinter) Date: Tue, 07 Jul 2009 13:36:27 +0200 Subject: nginx 0.7.61 keeps crashing Message-ID: <4A53333B.1020605@internetx.de> Hello, i got a nginx 0.7.61 running on centos 5.3 x86_64 which keeps crashing with panic: MUTEX_LOCK (22) [op.c:352]. any ideas? or is this a known problem? From phill at theactivitypeople.co.uk Tue Jul 7 15:51:41 2009 From: phill at theactivitypeople.co.uk (Phillip Oldham) Date: Tue, 07 Jul 2009 12:51:41 +0100 Subject: Maxmind GeoIP City Module Message-ID: <4A5336CD.9050308@theactivitypeople.co.uk> We're in need of a GeoIP module which works with the current development branch of nginx and can resolve an IP to City-level using Maxmind's GeoIP Database (the .dat files). We use nginx extensively but don't have experience in C to develop such a module in-house. Are there any module developers on the list who would be interested in such a project? We're flexible regarding the funding of development for such a module, and also flexible on whether the finshed module is open-sourced and released back to the community. If anyone is interested please feel free to mail me off-list with any thoughts/costs/etc. -- *Phillip B Oldham* ActivityHQ phill at activityhq.com ------------------------------------------------------------------------ *Policies* This e-mail and its attachments are intended for the above named recipient(s) only and may be confidential. If they have come to you in error, please reply to this e-mail and highlight the error. No action should be taken regarding content, nor must you copy or show them to anyone. This e-mail has been created in the knowledge that Internet e-mail is not a 100% secure communications medium, and we have taken steps to ensure that this e-mail and attachments are free from any virus. We must advise that in keeping with good computing practice the recipient should ensure they are completely virus free, and that you understand and observe the lack of security when e-mailing us. ------------------------------------------------------------------------ -------------- next part -------------- An HTML attachment was scrubbed... URL: From mdounin at mdounin.ru Tue Jul 7 16:18:02 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Tue, 7 Jul 2009 16:18:02 +0400 Subject: nginx 0.7.61 keeps crashing In-Reply-To: <4A53333B.1020605@internetx.de> References: <4A53333B.1020605@internetx.de> Message-ID: <20090707121802.GU5360@mdounin.ru> Hello! On Tue, Jul 07, 2009 at 01:36:27PM +0200, Juergen Gotteswinter wrote: > Hello, > > i got a nginx 0.7.61 running on centos 5.3 x86_64 which keeps crashing with > > panic: MUTEX_LOCK (22) [op.c:352]. > > any ideas? or is this a known problem? It's messaged issued by perl (for some not-yet-investigated reason). If you don't need embedded perl - just compile nginx without it. Am I right in the assumption that this happens on restart? If yes, there were some reports that fixing init script to wait for nginx to stop before starting a new copy (or at least adding a sleep 1 between stop and start in restart function) helps. Or better - you may use reload instead (since there are no real reason to restart nginx, reload usually sufficient and doesn't loose any single request). Maxim Dounin From jerome at loyet.net Tue Jul 7 16:19:59 2009 From: jerome at loyet.net (=?ISO-8859-1?B?Suly9G1lIExveWV0?=) Date: Tue, 7 Jul 2009 14:19:59 +0200 Subject: Thanks ! Message-ID: <3bea96c40907070519n1235af88sba7328478dfb96ce@mail.gmail.com> Hi all, Hi Igor, I've been working on a huge event for one of my customer. We were running apache / php5_module. We switched to nginx with proxy to apache / php5_module. The proxy_cache feature and gzip level 9 have been used. Last year for the same event we hadn't been able to handle the stream with 8 servers ... this year we've been able to handle it without any problems with only 2 servers (only one should have been enough but we needed redundancy). A summary: > 300 Mbits/s out bandwith > 60 Mbits/s in bandwith > 7000 requests / s < 60% cpu per server since 8AM more than 50 000 000 hits have been served response time constant and excellent (more than 90% of the requests have been served in less than 10ms) Nginx 0.8.4 (I know this is not a production version but It works so good. I don't have any good reasons not to use it.) I'm not suprise by thoses results as it's been benched recently. But I just want to thanks Igor and all of you for your time working on nginx and the amazing work accomplished. Few months back, the first time I looked at nginx, the cache support didn't even exist (except for proxy_store) and now it's fully working ... it was just all we were waiting for. I'll try to make an article somewhere about all that but I'm not sure I'll be able to (politic is politic). ++ Jerome From juergen.gotteswinter at internetx.de Tue Jul 7 16:22:24 2009 From: juergen.gotteswinter at internetx.de (=?ISO-8859-1?Q?InterNetX_-_J=FCrgen_Gotteswinter?=) Date: Tue, 07 Jul 2009 14:22:24 +0200 Subject: nginx 0.7.61 keeps crashing In-Reply-To: <4A53333B.1020605@internetx.de> References: <4A53333B.1020605@internetx.de> Message-ID: <4A533E00.4090006@internetx.de> i found some discussions about this, and it seems that the nginx perl module can cause this on systems with threaded perl. i recompiled the rpm without the "--with-perl..." part and reinstalled nginx. a few minutes later nginx died again, here are the last lines from error.log 2009/07/07 14:00:04 [notice] 15087#0: signal 3 (SIGQUIT) received, shutting down 2009/07/07 14:00:04 [notice] 15088#0: gracefully shutting down 2009/07/07 14:00:07 [notice] 15088#0: exiting 2009/07/07 14:00:07 [notice] 15088#0: exit 2009/07/07 14:00:07 [notice] 15087#0: signal 17 (SIGCHLD) received 2009/07/07 14:00:07 [notice] 15087#0: worker process 15088 exited with code 0 2009/07/07 14:00:07 [notice] 15087#0: exit any ideas? On 07/07/2009 01:36 PM, Juergen Gotteswinter wrote: > Hello, > > i got a nginx 0.7.61 running on centos 5.3 x86_64 which keeps crashing > with > > panic: MUTEX_LOCK (22) [op.c:352]. > > any ideas? or is this a known problem? > > > -- InterNetX GmbH Maximilianstrasse 6 D-93047 Regensburg Tel. +49 941 59559-480 Fax +49 941 59559-245 Gesch?ftsf?hrer/CEO: Thomas M?rz Amtsgericht Regensburg, HRB 7142 From mdounin at mdounin.ru Tue Jul 7 16:29:55 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Tue, 7 Jul 2009 16:29:55 +0400 Subject: nginx 0.7.61 keeps crashing In-Reply-To: <4A533E00.4090006@internetx.de> References: <4A53333B.1020605@internetx.de> <4A533E00.4090006@internetx.de> Message-ID: <20090707122954.GV5360@mdounin.ru> Hello! On Tue, Jul 07, 2009 at 02:22:24PM +0200, InterNetX - J?rgen Gotteswinter wrote: > i found some discussions about this, and it seems that the nginx perl > module can cause this on systems with threaded perl. i recompiled the > rpm without the "--with-perl..." part and reinstalled nginx. a few > minutes later nginx died again, here are the last lines from error.log > > 2009/07/07 14:00:04 [notice] 15087#0: signal 3 (SIGQUIT) received, > shutting down > 2009/07/07 14:00:04 [notice] 15088#0: gracefully shutting down > 2009/07/07 14:00:07 [notice] 15088#0: exiting > 2009/07/07 14:00:07 [notice] 15088#0: exit > 2009/07/07 14:00:07 [notice] 15087#0: signal 17 (SIGCHLD) received > 2009/07/07 14:00:07 [notice] 15087#0: worker process 15088 exited with > code 0 > 2009/07/07 14:00:07 [notice] 15087#0: exit > > any ideas? Somebody sent QUIT signal to nginx master. From the time (14:00) it looks like wrongly implemented log rotation (e.g. via restart or something like). For log rotation use USR1 signal, it was specially designated for this (and instructs nginx to just reopen logs without any extra unneeded work). Maxim Dounin > > On 07/07/2009 01:36 PM, Juergen Gotteswinter wrote: >> Hello, >> >> i got a nginx 0.7.61 running on centos 5.3 x86_64 which keeps crashing >> with >> >> panic: MUTEX_LOCK (22) [op.c:352]. >> >> any ideas? or is this a known problem? >> >> >> > > -- > InterNetX GmbH > Maximilianstrasse 6 > D-93047 Regensburg > > Tel. +49 941 59559-480 > Fax +49 941 59559-245 > > Gesch?ftsf?hrer/CEO: Thomas M?rz > Amtsgericht Regensburg, HRB 7142 > > From jemminger at gmail.com Tue Jul 7 16:46:58 2009 From: jemminger at gmail.com (jeff emminger) Date: Tue, 7 Jul 2009 08:46:58 -0400 Subject: Thanks ! In-Reply-To: <3bea96c40907070519n1235af88sba7328478dfb96ce@mail.gmail.com> References: <3bea96c40907070519n1235af88sba7328478dfb96ce@mail.gmail.com> Message-ID: <8b109c50907070546v7d50762ncffa1346e5a0c343@mail.gmail.com> 2009/7/7 J?r?me Loyet : > Hi all, Hi Igor, > > I've been working on a huge event for one of my customer. We were > running apache / php5_module. We switched to nginx with proxy to > apache / php5_module. The proxy_cache feature and gzip level 9 have > been used. > I'm interested, why does nginx just proxy back to apache+php? Is apache's balancing the inefficient part? From jerome at loyet.net Tue Jul 7 16:54:53 2009 From: jerome at loyet.net (=?ISO-8859-1?B?Suly9G1lIExveWV0?=) Date: Tue, 7 Jul 2009 14:54:53 +0200 Subject: Thanks ! In-Reply-To: <8b109c50907070546v7d50762ncffa1346e5a0c343@mail.gmail.com> References: <3bea96c40907070519n1235af88sba7328478dfb96ce@mail.gmail.com> <8b109c50907070546v7d50762ncffa1346e5a0c343@mail.gmail.com> Message-ID: <3bea96c40907070554k5e003b84g683af2619d760dd2@mail.gmail.com> 2009/7/7 jeff emminger : > 2009/7/7 J?r?me Loyet : >> Hi all, Hi Igor, >> >> I've been working on a huge event for one of my customer. We were >> running apache / php5_module. We switched to nginx with proxy to >> apache / php5_module. The proxy_cache feature and gzip level 9 have >> been used. >> > > I'm interested, why does nginx just proxy back to apache+php? because the php code has been tested only on apache and the customer don't want to try something else. A third party compagny handle de development of the application. We have a lots of constrains here. > Is apache's balancing the inefficient part? What do you mean by "apache's balancing" ? From juergen.gotteswinter at internetx.de Tue Jul 7 17:03:18 2009 From: juergen.gotteswinter at internetx.de (=?UTF-8?B?SW50ZXJOZXRYIC0gSsO8cmdlbiBHb3R0ZXN3aW50ZXI=?=) Date: Tue, 07 Jul 2009 15:03:18 +0200 Subject: nginx 0.7.61 keeps crashing In-Reply-To: <20090707122954.GV5360@mdounin.ru> References: <4A53333B.1020605@internetx.de> <4A533E00.4090006@internetx.de> <20090707122954.GV5360@mdounin.ru> Message-ID: <4A534796.1070701@internetx.de> Hello Maxim, logrotate is done with -USR Signal, anyway i disabled the nginx logrotate for the first. lets see if it helps. Thanks for help :) Juergen On 07/07/2009 02:29 PM, Maxim Dounin wrote: > Hello! > > On Tue, Jul 07, 2009 at 02:22:24PM +0200, InterNetX - J?rgen Gotteswinter wrote: > > >> i found some discussions about this, and it seems that the nginx perl >> module can cause this on systems with threaded perl. i recompiled the >> rpm without the "--with-perl..." part and reinstalled nginx. a few >> minutes later nginx died again, here are the last lines from error.log >> >> 2009/07/07 14:00:04 [notice] 15087#0: signal 3 (SIGQUIT) received, >> shutting down >> 2009/07/07 14:00:04 [notice] 15088#0: gracefully shutting down >> 2009/07/07 14:00:07 [notice] 15088#0: exiting >> 2009/07/07 14:00:07 [notice] 15088#0: exit >> 2009/07/07 14:00:07 [notice] 15087#0: signal 17 (SIGCHLD) received >> 2009/07/07 14:00:07 [notice] 15087#0: worker process 15088 exited with >> code 0 >> 2009/07/07 14:00:07 [notice] 15087#0: exit >> >> any ideas? >> > Somebody sent QUIT signal to nginx master. From the time (14:00) > it looks like wrongly implemented log rotation (e.g. via restart > or something like). For log rotation use USR1 signal, it was > specially designated for this (and instructs nginx to just reopen > logs without any extra unneeded work). > > Maxim Dounin > > >> On 07/07/2009 01:36 PM, Juergen Gotteswinter wrote: >> >>> Hello, >>> >>> i got a nginx 0.7.61 running on centos 5.3 x86_64 which keeps crashing >>> with >>> >>> panic: MUTEX_LOCK (22) [op.c:352]. >>> >>> any ideas? or is this a known problem? >>> >>> >>> >>> >> -- >> InterNetX GmbH >> Maximilianstrasse 6 >> D-93047 Regensburg >> >> Tel. +49 941 59559-480 >> Fax +49 941 59559-245 >> >> Gesch?ftsf?hrer/CEO: Thomas M?rz >> Amtsgericht Regensburg, HRB 7142 >> >> >> > > -- InterNetX GmbH Maximilianstrasse 6 D-93047 Regensburg Tel. +49 941 59559-480 Fax +49 941 59559-245 Gesch?ftsf?hrer/CEO: Thomas M?rz Amtsgericht Regensburg, HRB 7142 From jemminger at gmail.com Tue Jul 7 18:09:57 2009 From: jemminger at gmail.com (jeff emminger) Date: Tue, 7 Jul 2009 10:09:57 -0400 Subject: Thanks ! In-Reply-To: <3bea96c40907070554k5e003b84g683af2619d760dd2@mail.gmail.com> References: <3bea96c40907070519n1235af88sba7328478dfb96ce@mail.gmail.com> <8b109c50907070546v7d50762ncffa1346e5a0c343@mail.gmail.com> <3bea96c40907070554k5e003b84g683af2619d760dd2@mail.gmail.com> Message-ID: <8b109c50907070709x7eba794fh7b53003fc46eb055@mail.gmail.com> >> Is apache's balancing the inefficient part? > > What do you mean by "apache's balancing" ? > > mod_balancer From jerome at loyet.net Tue Jul 7 18:22:12 2009 From: jerome at loyet.net (=?ISO-8859-1?B?Suly9G1lIExveWV0?=) Date: Tue, 7 Jul 2009 16:22:12 +0200 Subject: Thanks ! In-Reply-To: <8b109c50907070709x7eba794fh7b53003fc46eb055@mail.gmail.com> References: <3bea96c40907070519n1235af88sba7328478dfb96ce@mail.gmail.com> <8b109c50907070546v7d50762ncffa1346e5a0c343@mail.gmail.com> <3bea96c40907070554k5e003b84g683af2619d760dd2@mail.gmail.com> <8b109c50907070709x7eba794fh7b53003fc46eb055@mail.gmail.com> Message-ID: <3bea96c40907070722o736c7a78hd55a186e04fc0c1e@mail.gmail.com> 2009/7/7 jeff emminger : >>> Is apache's balancing the inefficient part? >> >> What do you mean by "apache's balancing" ? >> >> > > mod_balancer We're not using mod_balancers, we have two dedicated servers running LVS. LVS load balances requests to nginx which sent the request to apache / php on localhost. From jemminger at gmail.com Tue Jul 7 19:01:19 2009 From: jemminger at gmail.com (jeff emminger) Date: Tue, 7 Jul 2009 11:01:19 -0400 Subject: Thanks ! In-Reply-To: <3bea96c40907070722o736c7a78hd55a186e04fc0c1e@mail.gmail.com> References: <3bea96c40907070519n1235af88sba7328478dfb96ce@mail.gmail.com> <8b109c50907070546v7d50762ncffa1346e5a0c343@mail.gmail.com> <3bea96c40907070554k5e003b84g683af2619d760dd2@mail.gmail.com> <8b109c50907070709x7eba794fh7b53003fc46eb055@mail.gmail.com> <3bea96c40907070722o736c7a78hd55a186e04fc0c1e@mail.gmail.com> Message-ID: <8b109c50907070801x2b10a42cla43547dc158d773@mail.gmail.com> >> >> mod_balancer > > We're not using mod_balancers, we have two dedicated servers running > LVS. LVS load balances requests to nginx which sent the request to > apache / php on localhost. > > I'm just curious as to why putting nginx in front of it all allows apache+php to handle the load, when it seems to be handling the same requests (for the php) as before. Unless there is a lot of static content that nginx is now handling? From denis at gostats.ru Tue Jul 7 19:26:07 2009 From: denis at gostats.ru (Denis F. Latypoff) Date: Tue, 7 Jul 2009 22:26:07 +0700 Subject: Thanks ! In-Reply-To: <8b109c50907070801x2b10a42cla43547dc158d773@mail.gmail.com> References: <3bea96c40907070519n1235af88sba7328478dfb96ce@mail.gmail.com> <8b109c50907070546v7d50762ncffa1346e5a0c343@mail.gmail.com> <3bea96c40907070554k5e003b84g683af2619d760dd2@mail.gmail.com> <8b109c50907070709x7eba794fh7b53003fc46eb055@mail.gmail.com> <3bea96c40907070722o736c7a78hd55a186e04fc0c1e@mail.gmail.com> <8b109c50907070801x2b10a42cla43547dc158d773@mail.gmail.com> Message-ID: <77175107.20090707222607@gostats.ru> Hello jeff, Tuesday, July 7, 2009, 10:01:19 PM, you wrote: >>> >>> mod_balancer >> >> We're not using mod_balancers, we have two dedicated servers running >> LVS. LVS load balances requests to nginx which sent the request to >> apache / php on localhost. >> >> > I'm just curious as to why putting nginx in front of it all allows > apache+php to handle the load, when it seems to be handling the same > requests (for the php) as before. Unless there is a lot of static > content that nginx is now handling? nginx great reduces apache load with the many slow clients. So apache processes (slots) are freed very fast with the nginx installed in front of. -- Best regards, Denis mailto:denis at gostats.ru From jerome at loyet.net Tue Jul 7 19:34:24 2009 From: jerome at loyet.net (=?ISO-8859-1?B?Suly9G1lIExveWV0?=) Date: Tue, 7 Jul 2009 17:34:24 +0200 Subject: Thanks ! In-Reply-To: <8b109c50907070801x2b10a42cla43547dc158d773@mail.gmail.com> References: <3bea96c40907070519n1235af88sba7328478dfb96ce@mail.gmail.com> <8b109c50907070546v7d50762ncffa1346e5a0c343@mail.gmail.com> <3bea96c40907070554k5e003b84g683af2619d760dd2@mail.gmail.com> <8b109c50907070709x7eba794fh7b53003fc46eb055@mail.gmail.com> <3bea96c40907070722o736c7a78hd55a186e04fc0c1e@mail.gmail.com> <8b109c50907070801x2b10a42cla43547dc158d773@mail.gmail.com> Message-ID: <3bea96c40907070834t5242faa7t418c89d2d0c55937@mail.gmail.com> 2009/7/7 jeff emminger : >>> >>> mod_balancer >> >> We're not using mod_balancers, we have two dedicated servers running >> LVS. LVS load balances requests to nginx which sent the request to >> apache / php on localhost. >> >> > > I'm just curious as to why putting nginx in front of it all allows > apache+php to handle the load, when it seems to be handling the same > requests (for the php) as before. ?Unless there is a lot of static > content that nginx is now handling? nginx as a reverse proxy handles the connections (keepalived) so apache does not have to spend time and ressources to handle them. THen nginx is caching PHP pages, so apache load is much much less than before (hundreds of process befire, only several now). From gabe at vtunnel.com Tue Jul 7 19:34:43 2009 From: gabe at vtunnel.com (Gabriel Ramuglia) Date: Tue, 7 Jul 2009 08:34:43 -0700 Subject: Thanks ! In-Reply-To: <8b109c50907070801x2b10a42cla43547dc158d773@mail.gmail.com> References: <3bea96c40907070519n1235af88sba7328478dfb96ce@mail.gmail.com> <8b109c50907070546v7d50762ncffa1346e5a0c343@mail.gmail.com> <3bea96c40907070554k5e003b84g683af2619d760dd2@mail.gmail.com> <8b109c50907070709x7eba794fh7b53003fc46eb055@mail.gmail.com> <3bea96c40907070722o736c7a78hd55a186e04fc0c1e@mail.gmail.com> <8b109c50907070801x2b10a42cla43547dc158d773@mail.gmail.com> Message-ID: <932ea6c90907070834h5546151m66710cf3ca5f3a53@mail.gmail.com> If he's pushing massive amounts of bandwidth, just having nginx in front can help free up the apache slots, as apache only has to stay open long enough to send its result to nginx, rather than stay open long enough to handle the entire connection for the end user. that could be a double edged sword / backfire however if the data being sent is larger than the size the apache slot would typically be, but if not it would be a help there. and it sounds from what was said that the cache module has helped quite a bit as well. if you are dealing with static content read from disk, then using nginx's single threaded architecture is especially better suited to that compared to using apache. 600megabit of streaming or downloads would just fall over and die with apache, especially if the per stream rates started to suffer, the number of apache slots to support that would soar. On Tue, Jul 7, 2009 at 8:01 AM, jeff emminger wrote: >>> >>> mod_balancer >> >> We're not using mod_balancers, we have two dedicated servers running >> LVS. LVS load balances requests to nginx which sent the request to >> apache / php on localhost. >> >> > > I'm just curious as to why putting nginx in front of it all allows > apache+php to handle the load, when it seems to be handling the same > requests (for the php) as before. ?Unless there is a lot of static > content that nginx is now handling? > > From yanleech at gmail.com Tue Jul 7 21:47:34 2009 From: yanleech at gmail.com (Leechael Yim) Date: Wed, 8 Jul 2009 01:47:34 +0800 Subject: 404 instead of "no inout file specified"? In-Reply-To: <20090701142046.GG5360@mdounin.ru> References: <6b1504c40907010652ib89ca65j40f918eb4035c3d5@mail.gmail.com> <20090701142046.GG5360@mdounin.ru> Message-ID:


if (-f $request_filename) {    fastcgi_pass 127.0.0.1:9000;
}

OR:


if (!-f $request_filename) {
    return 404;
}
fastcgi_pass 127.0.0.1:9000;

The first one is what I current used with my Nginx. On Wed, Jul 1, 2009 at 10:20 PM, Maxim Dounin wrote: > Hello! > > On Wed, Jul 01, 2009 at 02:52:06PM +0100, Nuno Magalh?es wrote: > > > Greetings, > > > > If i provide a valid URL, it's processed. > > If i provide an invalid URL, nginx shows the 404 page except it the > > invalid URL ends in .php - in that case i get the "no input file > > specified" message. > > Looks like you've configured passing urls ending with ".php" to > php for processing. So you got errors from php, not from nginx. > It's expected behaviour, no? > > And, actually, "no input file specified" *is* 404 error. It's > just has some fancy php-specific content. > > > I'm in the process of changing my php files to an html extension (i > > assume that would "fix" things), however, i'd like to get a 404 > > instead of the message regardless. > > > > I tried using "fastcgi_intercept_errors off" in the conf for my server > > (in sites-available), but it didn't do anything (i did restart both > > nginx and the fast-cgi processes). > > > > Is there a better way? > > If you want to change error page returned by php to something > pretty-looking, you should: > > 1. Use "fastcgi_intercept_errors on;" to instruct nginx to > intercept and change errors returned by fastcgi application. > > 2. Make sure you explicitly configured error_page for 404. > Default compiled-in error pages are used only for nginx own > errors, not for fastcgi_intercept_errors/proxy_intercept_errors. > > So your resulting config should look like: > > error_page 404 /404.html; > > location ~ \.php$ { > fastcgi_pass ...; > fastcgi_intercept_errors on; > } > > ... > > Maxim Dounin > > -- Twitter: http://www.twitter.com/leechael -------------- next part -------------- An HTML attachment was scrubbed... URL: From edhoprima at gmail.com Tue Jul 7 22:09:50 2009 From: edhoprima at gmail.com (Edho P Arief) Date: Wed, 8 Jul 2009 01:09:50 +0700 Subject: 404 instead of "no inout file specified"? In-Reply-To: References: <6b1504c40907010652ib89ca65j40f918eb4035c3d5@mail.gmail.com> <20090701142046.GG5360@mdounin.ru> Message-ID: On Wed, Jul 8, 2009 at 12:47 AM, Leechael Yim wrote: >


> if (-f $request_filename) {
> ?? ?fastcgi_pass 127.0.0.1:9000;
> }
>

> OR: >


> if (!-f $request_filename) {
> ?? ?return 404;
> }
> fastcgi_pass 127.0.0.1:9000;
>

> The first one is what I current used with my Nginx. location ~ \.php$ { try_files $uri @404; ... } location @404 { return 404; # or custom error page } when will people stop using !-f $request_filename... -- O< ascii ribbon campaign - stop html mail - www.asciiribbon.org From nginx-forum at nginx.us Tue Jul 7 22:10:21 2009 From: nginx-forum at nginx.us (vburshteyn) Date: Tue, 7 Jul 2009 14:10:21 -0400 Subject: proxy_cache Message-ID: <2171f33932c1ad149e34661339012770.NginxMailingList@forum.nginx.org> Hi folks, Before i get started i read that somebody posted simmilar thread but since i am still learning this, the answer/solution did not make sense. I have set up proxy_cache on my nginx but it seems that it is not writing to the cache directory. http { upstream backend { server 10.176.67.151:8080; } proxy_cache_path /etc/nginx/cache/ levels=1:2 keys_zone=one:10m max_size=200m; #proxy_temp_path /etc/nginx/temp; server { listen 10.176.67.151:80; location / { proxy_pass http://backend; proxy_cache one; proxy_cache_key backend$request_uri; proxy_cache_valid 10m; include /etc/nginx/proxy.conf; } Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3826,3826#msg-3826 From pchychi at gmail.com Tue Jul 7 22:27:26 2009 From: pchychi at gmail.com (Payam Chychi) Date: Tue, 7 Jul 2009 11:27:26 -0700 Subject: proxy_cache In-Reply-To: <2171f33932c1ad149e34661339012770.NginxMailingList@forum.nginx.org> References: <2171f33932c1ad149e34661339012770.NginxMailingList@forum.nginx.org> Message-ID: On Tue, Jul 7, 2009 at 11:10 AM, vburshteyn wrote: > Hi folks, ?Before i get started i read that somebody posted simmilar thread but since i am still learning this, the answer/solution did not make sense. > > I have set up proxy_cache on my nginx but it seems that it is not writing to the cache directory. > > > http { > > upstream backend > { > server 10.176.67.151:8080; > } > > > proxy_cache_path /etc/nginx/cache/ levels=1:2 keys_zone=one:10m max_size=200m; > #proxy_temp_path /etc/nginx/temp; > > > server { > ? ?listen 10.176.67.151:80; > > location / { > proxy_pass http://backend; > proxy_cache one; > proxy_cache_key ? ? ? ? backend$request_uri; > proxy_cache_valid ? ? ? 10m; > include /etc/nginx/proxy.conf; > > } > > Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3826,3826#msg-3826 > > > can you run logging with debug? then show us the log file... it will show exactly what is happening. Also, i would recomend you NOT using the following if you host more than 1 site on teh same upstream: " proxy_cache_key backend$request_uri;" make it uniqu to every site if possible else you may get wrong content served back out to requests that have the same md5 -Payam -- Payam Tarverdyan Chychi Network Security Specialist / Network Engineer From yanleech at gmail.com Tue Jul 7 22:30:10 2009 From: yanleech at gmail.com (Leechael Yim) Date: Wed, 8 Jul 2009 02:30:10 +0800 Subject: 404 instead of "no inout file specified"? In-Reply-To: References: <6b1504c40907010652ib89ca65j40f918eb4035c3d5@mail.gmail.com> <20090701142046.GG5360@mdounin.ru> Message-ID: Some situation like, this moment, I got your email.... On Wed, Jul 8, 2009 at 2:09 AM, Edho P Arief wrote: > On Wed, Jul 8, 2009 at 12:47 AM, Leechael Yim wrote: > >


> > if (-f $request_filename) {
> >     fastcgi_pass 127.0.0.1:9000;
> > }
> >

> > OR: > >


> > if (!-f $request_filename) {
> >     return 404;
> > }
> > fastcgi_pass 127.0.0.1:9000;
> >

> > The first one is what I current used with my Nginx. > > > location ~ \.php$ { > try_files $uri @404; > ... > } > > location @404 { > return 404; # or custom error page > } > > when will people stop using !-f $request_filename... > > -- > O< ascii ribbon campaign - stop html mail - www.asciiribbon.org > > -- Twitter: http://www.twitter.com/leechael -------------- next part -------------- An HTML attachment was scrubbed... URL: From mdounin at mdounin.ru Tue Jul 7 23:38:52 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Tue, 7 Jul 2009 23:38:52 +0400 Subject: 404 instead of "no inout file specified"? In-Reply-To: References: <6b1504c40907010652ib89ca65j40f918eb4035c3d5@mail.gmail.com> <20090701142046.GG5360@mdounin.ru> Message-ID: <20090707193852.GA54083@mdounin.ru> Hello! On Wed, Jul 08, 2009 at 01:47:34AM +0800, Leechael Yim wrote: >


> if (-f $request_filename) {    fastcgi_pass 127.0.0.1:9000;
> }
>

> > OR: > >


> if (!-f $request_filename) {
>     return 404;
> }
> fastcgi_pass 127.0.0.1:9000;
>

> > The first one is what I current used with my Nginx. Both snippets 1. Use if() which is evil. It was explained a lot of times here why if's are evil, so I'm not going to repeat. 2. Just won't work with backend(s) on other host(s). Maxim Dounin > > On Wed, Jul 1, 2009 at 10:20 PM, Maxim Dounin wrote: > > > Hello! > > > > On Wed, Jul 01, 2009 at 02:52:06PM +0100, Nuno Magalh?es wrote: > > > > > Greetings, > > > > > > If i provide a valid URL, it's processed. > > > If i provide an invalid URL, nginx shows the 404 page except it the > > > invalid URL ends in .php - in that case i get the "no input file > > > specified" message. > > > > Looks like you've configured passing urls ending with ".php" to > > php for processing. So you got errors from php, not from nginx. > > It's expected behaviour, no? > > > > And, actually, "no input file specified" *is* 404 error. It's > > just has some fancy php-specific content. > > > > > I'm in the process of changing my php files to an html extension (i > > > assume that would "fix" things), however, i'd like to get a 404 > > > instead of the message regardless. > > > > > > I tried using "fastcgi_intercept_errors off" in the conf for my server > > > (in sites-available), but it didn't do anything (i did restart both > > > nginx and the fast-cgi processes). > > > > > > Is there a better way? > > > > If you want to change error page returned by php to something > > pretty-looking, you should: > > > > 1. Use "fastcgi_intercept_errors on;" to instruct nginx to > > intercept and change errors returned by fastcgi application. > > > > 2. Make sure you explicitly configured error_page for 404. > > Default compiled-in error pages are used only for nginx own > > errors, not for fastcgi_intercept_errors/proxy_intercept_errors. > > > > So your resulting config should look like: > > > > error_page 404 /404.html; > > > > location ~ \.php$ { > > fastcgi_pass ...; > > fastcgi_intercept_errors on; > > } > > > > ... > > > > Maxim Dounin > > > > > > > -- > Twitter: http://www.twitter.com/leechael From tlott at gamesnet.de Wed Jul 8 01:52:00 2009 From: tlott at gamesnet.de (Tobias Lott) Date: Tue, 7 Jul 2009 23:52:00 +0200 Subject: Did I got try_files wrong? and bad gateway 502 In-Reply-To: <20090707020609.GS5360@mdounin.ru> References: <20090706053727.43700b5d@sub.han.vpn.gamesnet.de> <20090706120341.5ab5e64f@sub.han.vpn.gamesnet.de> <20090706235202.7b12b96d@sub.han.vpn.gamesnet.de> <20090707020609.GS5360@mdounin.ru> Message-ID: <20090707235200.718e323c@sub.han.vpn.gamesnet.de> On Tue, 7 Jul 2009 06:06:10 +0400 Maxim Dounin wrote: > Hello! > > On Mon, Jul 06, 2009 at 11:52:02PM +0200, Tobias Lott wrote: > > > > > > > On Mon, 6 Jul 2009 13:29:58 -0700 > > Michael Shadle wrote: > > > > > Check the nginx error log for what the message is when you hit > > > those 502's. > > > > > > Perhaps apache is too busy to service the request or is sending > > > malformed resposne etc? > > > > > > On Mon, Jul 6, 2009 at 3:03 AM, Tobias Lott > > > wrote: > > > > > > > > > > > > On Sun, 5 Jul 2009 21:39:44 -0700 > > > > Michael Shadle wrote: > > > > > > > >> On Sun, Jul 5, 2009 at 8:37 PM, Tobias Lott > > > >> wrote: > > > >> > > > >> > Additionally I got another issue, which is that customers > > > >> > getting 502s pretty damn randomly (using nginx-0.7.61 as > > > >> > frontend and apache 2.2.11 as backend for the time being, > > > >> > which gonna gets migrated soon). Was trying to tweak a lil > > > >> > with proxy_*_timeout and keepalive_timeout but since its not > > > >> > reproducible I can't be sure its fixed, are there some "good > > > >> > to go" values or another option to check? > > > >> > > > >> Is it nginx giving the 502 or apache? > > > >> > > > > > > > > nginx is giving the 502 > > > > > > > > -- > > > > Tobias Lott > > > > > > > > > > > > > > > The only error log entry thats consistent with the 502 is the > > following, but its not the same file, one type of file, same client: > > > > 2009/07/06 22:48:20 [crit] 86851#0: *183664 connect() to > > XX.XX.XX.XX:XX failed (1: Operation not permitted) while connecting > > to upstream, client: XX.XX.XX.XX, server: hostname.tld, request: > > "GET /some/random/picture.jpg HTTP/1.1", upstream: > > "http://XX.XX.XX.XX:XX/some/random/picture.jpg", > > host: "hostname.tld", referrer: "http://hostname.tld/some/" > > This is EPERM, and most typical reasons include statefull firewall > between nginx frontend and backend or timewait reuse switched off. > > If it's statefull firewall - remove it or convert to non-statefull > one. > > If it's timewait issue - switch on timewait reuse/recycle (see your > OS tuning guides to find out how). > > > There is nothing in Apache error/access log at that time. > > Since nginx wasn't allowed to connect() - it's somewhat expected. > > Maxim Dounin > True TIME_WAIT is the Problem, thanks for the hint! Anybody got an idea to my other Issue?: >I got a problem understanding try_files I guess. > >Got this as configuration : > >location = / { rewrite ^(.*)$ /shop/ redirect; } > >location / { > proxy_pass http://backend:10060; > proxy_set_header X-Real-IP $remote_addr; > proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for; > proxy_set_header Host $http_host; > proxy_redirect off; > proxy_connect_timeout 90; > proxy_send_timeout 90; > proxy_read_timeout 120; >} > > >Problem is that I cant "catch" all 404 and rewrite it >to /shop/index.php but with "if" (which I shouldn't use cause of >possible performance issues, Igor mentioned I think.) > >Doesn't work: >try_files $uri $uri/ /shop/index.php; >error_page 404 /shop/index.php; > >Works: >if (!-f $request_filename) { rewrite ^(.*)$ /shop/index.php >redirect; } -- Tobias Lott From merlincorey at dc949.org Wed Jul 8 02:02:56 2009 From: merlincorey at dc949.org (merlin corey) Date: Tue, 7 Jul 2009 15:02:56 -0700 Subject: try_files and static web site In-Reply-To: <53F14C2F-996A-4EAB-B0CF-AC209C2B01E0@gmail.com> References: <68dea8f82aef5a4f9d2f1d4e02a49f5b@ruby-forum.com> <53F14C2F-996A-4EAB-B0CF-AC209C2B01E0@gmail.com> Message-ID: Agreed. There is nothing it is doing, no fallback location (always should be your last location in try_files). On Mon, Jul 6, 2009 at 10:47 PM, Michael Shadle wrote: > I see no reason to be using try_files at all in your config. > > Sent from my iPhone > > On Jul 6, 2009, at 10:33 PM, Jauder Ho wrote: > >> I have a site that is completely static and I figured I would use >> try_files however this results in HTTP 500 errors for directories i.e. >> trying to get to http://foo.com/bar/ results in a 500 error. What's >> wrong with this config? Thanks. >> >> --Jauder >> >> ?server { >> ? # port to listen on. Can also be set to an IP:PORT >> ? listen 80; >> >> ? # Set the charset >> ? charset utf-8; >> >> ? # sets the domain[s] that this vhost server requests for >> ? server_name www.foo.com; >> >> ? # doc root >> ? root /var/www/foo.com; >> >> ? # vhost specific access log >> ? access_log ?/var/log/nginx_access.log ?main; >> >> >> ? # Catchall for everything else >> ? location / { >> ? ? root /var/www/foo.com; >> ? ? access_log off; >> >> ? ? index index.html index.shtml; >> ? ? expires 1d; >> >> ? ? try_files $uri $uri/; >> >> ? ? #if (-f $request_filename) { >> ? ? # ?break; >> ? ? #} >> ? } >> >> >> >> ? error_page 500 502 503 504 /500.html; >> ? error_page 404 /404.html; >> >> ?} >> -- >> Posted via http://www.ruby-forum.com/. >> > > From nginx-forum at nginx.us Wed Jul 8 07:26:54 2009 From: nginx-forum at nginx.us (davidc) Date: Tue, 7 Jul 2009 23:26:54 -0400 Subject: Logging embedded Perl return value In-Reply-To: References: Message-ID: <5c3917db97182ee1a2fa973d65d5dd3e.NginxMailingList@forum.nginx.org> I tried patching nginx to produce a '210 Custom Response' which is just a 2xx success response but without any content. As it doesn't use content I thought it should behave like the existing HTTP_NO_CONTENT response. Is there any problem in doing it this way? Is there a better way to add it? diff -crB /usr/src/redhat/BUILD/nginx-0.7.61/src/http/modules/ngx_http_headers_filter_module.c nginx-0.7.61/src/http/modules/ngx_http_headers_filter_module.c *** /usr/src/redhat/BUILD/nginx-0.7.61/src/http/modules/ngx_http_headers_filter_module.c 2009-03-26 21:34:37.000000000 +0800 --- nginx-0.7.61/src/http/modules/ngx_http_headers_filter_module.c 2009-07-08 10:25:54.000000000 +0800 *************** *** 145,150 **** --- 145,151 ---- || r != r->main || (r->headers_out.status != NGX_HTTP_OK && r->headers_out.status != NGX_HTTP_NO_CONTENT + && r->headers_out.status != NGX_HTTP_CUSTOM_RESPONSE && r->headers_out.status != NGX_HTTP_MOVED_PERMANENTLY && r->headers_out.status != NGX_HTTP_MOVED_TEMPORARILY && r->headers_out.status != NGX_HTTP_NOT_MODIFIED)) diff -crB /usr/src/redhat/BUILD/nginx-0.7.61/src/http/modules/perl/nginx.pm nginx-0.7.61/src/http/modules/perl/nginx.pm *** /usr/src/redhat/BUILD/nginx-0.7.61/src/http/modules/perl/nginx.pm 2009-06-22 17:23:35.000000000 +0800 --- nginx-0.7.61/src/http/modules/perl/nginx.pm 2009-07-08 10:23:53.000000000 +0800 *************** *** 16,21 **** --- 16,22 ---- HTTP_CREATED HTTP_NO_CONTENT HTTP_PARTIAL_CONTENT + HTTP_CUSTOM_RESPONSE HTTP_MOVED_PERMANENTLY HTTP_MOVED_TEMPORARILY *************** *** 61,66 **** --- 62,68 ---- use constant HTTP_CREATED => 201; use constant HTTP_NO_CONTENT => 204; use constant HTTP_PARTIAL_CONTENT => 206; + use constant HTTP_CUSTOM_RESPONSE => 210; use constant HTTP_MOVED_PERMANENTLY => 301; use constant HTTP_MOVED_TEMPORARILY => 302; diff -crB /usr/src/redhat/BUILD/nginx-0.7.61/src/http/ngx_http_header_filter_module.c nginx-0.7.61/src/http/ngx_http_header_filter_module.c *** /usr/src/redhat/BUILD/nginx-0.7.61/src/http/ngx_http_header_filter_module.c 2009-05-18 20:58:19.000000000 +0800 --- nginx-0.7.61/src/http/ngx_http_header_filter_module.c 2009-07-08 10:36:55.000000000 +0800 *************** *** 58,67 **** ngx_string("204 No Content"), ngx_null_string, /* "205 Reset Content" */ ngx_string("206 Partial Content"), /* ngx_null_string, */ /* "207 Multi-Status" */ ! #define NGX_HTTP_LAST_LEVEL_200 207 #define NGX_HTTP_LEVEL_200 (NGX_HTTP_LAST_LEVEL_200 - 200) /* ngx_null_string, */ /* "300 Multiple Choices" */ --- 58,71 ---- ngx_string("204 No Content"), ngx_null_string, /* "205 Reset Content" */ ngx_string("206 Partial Content"), + ngx_null_string, + ngx_null_string, + ngx_null_string, + ngx_string("210 Custom Response"), /* ngx_null_string, */ /* "207 Multi-Status" */ ! #define NGX_HTTP_LAST_LEVEL_200 211 #define NGX_HTTP_LEVEL_200 (NGX_HTTP_LAST_LEVEL_200 - 200) /* ngx_null_string, */ /* "300 Multiple Choices" */ *************** *** 220,226 **** { /* 2XX */ ! if (status == NGX_HTTP_NO_CONTENT) { r->header_only = 1; r->headers_out.content_type.len = 0; r->headers_out.content_type.data = NULL; --- 224,230 ---- { /* 2XX */ ! if (status == NGX_HTTP_NO_CONTENT || status == NGX_HTTP_CUSTOM_RESPONSE) { r->header_only = 1; r->headers_out.content_type.len = 0; r->headers_out.content_type.data = NULL; diff -crB /usr/src/redhat/BUILD/nginx-0.7.61/src/http/ngx_http_request.c nginx-0.7.61/src/http/ngx_http_request.c *** /usr/src/redhat/BUILD/nginx-0.7.61/src/http/ngx_http_request.c 2009-06-22 17:31:33.000000000 +0800 --- nginx-0.7.61/src/http/ngx_http_request.c 2009-07-08 10:23:21.000000000 +0800 *************** *** 1852,1858 **** if (rc >= NGX_HTTP_SPECIAL_RESPONSE || rc == NGX_HTTP_CREATED ! || rc == NGX_HTTP_NO_CONTENT) { if (rc == NGX_HTTP_CLOSE) { ngx_http_close_request(r, rc); --- 1852,1859 ---- if (rc >= NGX_HTTP_SPECIAL_RESPONSE || rc == NGX_HTTP_CREATED ! || rc == NGX_HTTP_NO_CONTENT ! || rc == NGX_HTTP_CUSTOM_RESPONSE) { if (rc == NGX_HTTP_CLOSE) { ngx_http_close_request(r, rc); diff -crB /usr/src/redhat/BUILD/nginx-0.7.61/src/http/ngx_http_request.h nginx-0.7.61/src/http/ngx_http_request.h *** /usr/src/redhat/BUILD/nginx-0.7.61/src/http/ngx_http_request.h 2009-05-22 19:05:26.000000000 +0800 --- nginx-0.7.61/src/http/ngx_http_request.h 2009-07-08 10:22:48.000000000 +0800 *************** *** 67,72 **** --- 67,74 ---- #define NGX_HTTP_NO_CONTENT 204 #define NGX_HTTP_PARTIAL_CONTENT 206 + #define NGX_HTTP_CUSTOM_RESPONSE 210 + #define NGX_HTTP_SPECIAL_RESPONSE 300 #define NGX_HTTP_MOVED_PERMANENTLY 301 #define NGX_HTTP_MOVED_TEMPORARILY 302 diff -crB /usr/src/redhat/BUILD/nginx-0.7.61/src/http/ngx_http_script.c nginx-0.7.61/src/http/ngx_http_script.c *** /usr/src/redhat/BUILD/nginx-0.7.61/src/http/ngx_http_script.c 2009-04-27 19:32:33.000000000 +0800 --- nginx-0.7.61/src/http/ngx_http_script.c 2009-07-08 10:27:24.000000000 +0800 *************** *** 1281,1287 **** e->status = code->status; ! if (code->status == NGX_HTTP_NO_CONTENT) { e->request->header_only = 1; e->request->zero_body = 1; } --- 1281,1289 ---- e->status = code->status; ! if (code->status == NGX_HTTP_NO_CONTENT ! || code->status == NGX_HTTP_CUSTOM_RESPONSE) { ! e->request->header_only = 1; e->request->zero_body = 1; } diff -crB /usr/src/redhat/BUILD/nginx-0.7.61/src/http/ngx_http_special_response.c nginx-0.7.61/src/http/ngx_http_special_response.c *** /usr/src/redhat/BUILD/nginx-0.7.61/src/http/ngx_http_special_response.c 2009-05-25 17:06:29.000000000 +0800 --- nginx-0.7.61/src/http/ngx_http_special_response.c 2009-07-08 10:28:09.000000000 +0800 *************** *** 404,410 **** err = 0; r->header_only = 1; ! } else if (error == NGX_HTTP_NO_CONTENT) { /* 204 */ err = 0; --- 404,410 ---- err = 0; r->header_only = 1; ! } else if (error == NGX_HTTP_NO_CONTENT || error == NGX_HTTP_CUSTOM_RESPONSE) { /* 204 */ err = 0; Posted at Nginx Forum: http://forum.nginx.org/read.php?2,933,3857#msg-3857 From jg at internetx.de Wed Jul 8 12:39:11 2009 From: jg at internetx.de (Juergen Gotteswinter) Date: Wed, 08 Jul 2009 10:39:11 +0200 Subject: Critical SSL Errors Message-ID: <4A545B2F.1040506@internetx.de> Hallo, i get those critical ssl errors in our nginx error log 2009/07/08 10:33:35 [crit] 15582#0: *382780 SSL_do_handshake() failed (SSL: error:140760FC:SSL routines:SSL23_GET_CLIENT_HELLO:unknown protocol) while SSL handshaking, client: 212.19.39.200, server: XXXXXXXXXXX the ssl connections cant be established in this case. anyone seen something like this before? thanks Juergen From is at rambler-co.ru Wed Jul 8 15:39:27 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Wed, 8 Jul 2009 15:39:27 +0400 Subject: Critical SSL Errors In-Reply-To: <4A545B2F.1040506@internetx.de> References: <4A545B2F.1040506@internetx.de> Message-ID: <20090708113927.GE49412@rambler-co.ru> On Wed, Jul 08, 2009 at 10:39:11AM +0200, Juergen Gotteswinter wrote: > Hallo, > > i get those critical ssl errors in our nginx error log > > 2009/07/08 10:33:35 [crit] 15582#0: *382780 SSL_do_handshake() failed > (SSL: error:140760FC:SSL routines:SSL23_GET_CLIENT_HELLO:unknown > protocol) while SSL handshaking, client: 212.19.39.200, server: XXXXXXXXXXX > > the ssl connections cant be established in this case. anyone seen > something like this before? The most SSL errors were logged on crit level until recently. Now at least in 0.7.x many of them are logged on info level. -- Igor Sysoev http://sysoev.ru/en/ From is at rambler-co.ru Wed Jul 8 17:15:19 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Wed, 8 Jul 2009 17:15:19 +0400 Subject: Nginx Punycode Support In-Reply-To: <4A51B59A.9030208@internetx.de> References: <4A51B59A.9030208@internetx.de> Message-ID: <20090708131519.GL49412@rambler-co.ru> On Mon, Jul 06, 2009 at 10:28:10AM +0200, Juergen Gotteswinter wrote: > Hello > > does nginx support punycode domains in the server_name directives? If you mean a domain like this: server_name xn--jrgen-gotteswinter-m6b.de; then it's possible. -- Igor Sysoev http://sysoev.ru/en/ From nginx-forum at nginx.us Wed Jul 8 17:18:24 2009 From: nginx-forum at nginx.us (vburshteyn) Date: Wed, 8 Jul 2009 09:18:24 -0400 Subject: vburshteyn Message-ID: <2c449df47c612bdeaa42d965f6c610af.NginxMailingList@forum.nginx.org> Hi folks. back again for more info :) I need to use the rewrite module to rewrite redirects. for example http://www.mysite.com/9-to-5/group_show/569889 needs to be http://groups.bqa.mysite.com/9-to-5/group_show/569889 any idea how i would go about doing this? Thanks in advance. Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3863,3863#msg-3863 From is at rambler-co.ru Wed Jul 8 17:33:37 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Wed, 8 Jul 2009 17:33:37 +0400 Subject: Did I got try_files wrong? and bad gateway 502 In-Reply-To: <20090706053727.43700b5d@sub.han.vpn.gamesnet.de> References: <20090706053727.43700b5d@sub.han.vpn.gamesnet.de> Message-ID: <20090708133337.GM49412@rambler-co.ru> On Mon, Jul 06, 2009 at 05:37:27AM +0200, Tobias Lott wrote: > Hey Folks > > I got a problem understanding try_files I guess. > > Got this as configuration : > > location = / { rewrite ^(.*)$ /shop/ redirect; } > > location / { > proxy_pass http://backend:10060; > proxy_set_header X-Real-IP $remote_addr; > proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for; > proxy_set_header Host $http_host; > proxy_redirect off; > proxy_connect_timeout 90; > proxy_send_timeout 90; > proxy_read_timeout 120; > } > > > Problem is that I cant "catch" all 404 and rewrite it > to /shop/index.php but with "if" (which I shouldn't use cause of > possible performance issues, Igor mentioned I think.) > > Doesn't work: > try_files $uri $uri/ /shop/index.php; > error_page 404 /shop/index.php; > > Works: > if (!-f $request_filename) { rewrite ^(.*)$ /shop/index.php > redirect; } Have you tried this: location = / { rewrite ^ /shop/ redirect; } location / { try_files $uri $uri/ /shop/index.php; proxy_pass http://backend:10060; ... } -- Igor Sysoev http://sysoev.ru/en/ From nginx-forum at nginx.us Wed Jul 8 18:15:03 2009 From: nginx-forum at nginx.us (ygo) Date: Wed, 8 Jul 2009 10:15:03 -0400 Subject: nginx API docs in english Message-ID: Is there a chance to see this page http://www.grid.net.ru/nginx/nginx-modules.html translated in English (by a human :) ) ? I am trying to understand nginx and automatic translation is sometime weird and doesn't help at all. Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3867,3867#msg-3867 From marlon at hyves.nl Wed Jul 8 18:17:55 2009 From: marlon at hyves.nl (Marlon de Boer) Date: Wed, 08 Jul 2009 16:17:55 +0200 Subject: syslog support for the 0.7 and 0.8 branch Message-ID: <4A54AA93.7040302@hyves.nl> Hi list, Had some private emails about a new patch, it can be downloaded at http://bugs.gentoo.org/attachment.cgi?id=197180. Thanks to Ben Taitelbaum for making it 0.7 / 0.8 compatible. Regards Marlon From valery+nginxen at grid.net.ru Wed Jul 8 18:56:55 2009 From: valery+nginxen at grid.net.ru (Valery Kholodkov) Date: Wed, 8 Jul 2009 15:56:55 +0100 (BST) Subject: nginx API docs in english In-Reply-To: Message-ID: <23257796.82771247065015837.JavaMail.root@zone.mtgsy.net> It is part of my plan to translate this article to English. May be some day in the future. ----- ygo wrote: > Is there a chance to see this page http://www.grid.net.ru/nginx/nginx-modules.html translated in English (by a human :) ) ? I am trying to understand nginx and automatic translation is sometime weird and doesn't help at all. -- Regards, Valery Kholodkov From eden at mojiti.com Wed Jul 8 19:29:32 2009 From: eden at mojiti.com (Eden Li) Date: Wed, 8 Jul 2009 08:29:32 -0700 Subject: nginx API docs in english In-Reply-To: References: Message-ID: <180e6a10907080829o5a7459can131e9846a7c33a64@mail.gmail.com> I don't think it's a direct translation, but this guide is *very* helpful: http://www.evanmiller.org/nginx-modules-guide.html I think it was written when 0.6.x was just becoming stable, but most of it should still apply to 0.7.x On Wed, Jul 8, 2009 at 7:15 AM, ygo wrote: > Is there a chance to see this page http://www.grid.net.ru/nginx/nginx-modules.html translated in English (by a human :) ) ? I am trying to understand nginx and automatic translation is sometime weird and doesn't help at all. > > Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3867,3867#msg-3867 > > > From cliff at develix.com Thu Jul 9 01:14:06 2009 From: cliff at develix.com (Cliff Wells) Date: Wed, 08 Jul 2009 14:14:06 -0700 Subject: Nginx proxying WebDAV = no love Message-ID: <1247087646.3689.9.camel@portable-evil> I'm experimenting with WebDAV using pywebdav: http://code.google.com/p/pywebdav/ It appears to work quite well if I connect to it directly (using either Nautilus and Cadaver as client). However, when I attempt to proxy to it via Nginx, I get messages that it's not a WebDAV-enabled share. Here's my nginx.conf: http { include mime.types; default_type application/octet-stream; server { listen 80; server_name _; location / { proxy_pass_header Server; proxy_set_header Host $http_host; proxy_set_header X-Forwarded-For $remote_addr; proxy_pass http://127.0.0.1:8008; } } } Cadaver reports the following (nautilus message isn't too informative): $ cadaver http://localhost/ Could not access / (not WebDAV-enabled?): Unknown transfer-coding in response Connection to `localhost' closed. dav:!> I can see the request reaches pywebdav: localhost - - [08/Jul/2009 14:10:40] "OPTIONS / HTTP/1.0" 200 - localhost - - [08/Jul/2009 14:10:40] "PROPFIND / HTTP/1.0" 207 - Nginx access log shows: 127.0.0.1 - - [08/Jul/2009:14:11:20 -0700] "OPTIONS / HTTP/1.1" 200 0 "-" "cadaver/0.23.2 neon/0.28.2" 127.0.0.1 - - [08/Jul/2009:14:11:20 -0700] "PROPFIND / HTTP/1.1" 207 593 "-" "cadaver/0.23.2 neon/0.28.2" Nginx error log shows nothing. Regards, Cliff -- http://www.google.com/search?q=vonage+sucks From mike503 at gmail.com Thu Jul 9 02:08:40 2009 From: mike503 at gmail.com (Michael Shadle) Date: Wed, 8 Jul 2009 15:08:40 -0700 Subject: Nginx proxying WebDAV = no love In-Reply-To: <1247087646.3689.9.camel@portable-evil> References: <1247087646.3689.9.camel@portable-evil> Message-ID: Does nginx even support propfind? Or options? I see it in the code, but I don't see anything that happens when someone issues the verb. It's almost like it stubbed out for future expansion (or so it doesn't reply with "invalid request" at least) On Wed, Jul 8, 2009 at 2:14 PM, Cliff Wells wrote: > I'm experimenting with WebDAV using pywebdav: > > http://code.google.com/p/pywebdav/ > > It appears to work quite well if I connect to it > directly (using either Nautilus and Cadaver as client). > > However, when I attempt to proxy to it via Nginx, > I get messages that it's not a WebDAV-enabled share. > > Here's my nginx.conf: > > http { > ? ?include ? ? ? mime.types; > ? ?default_type ?application/octet-stream; > > ? ?server { > ? ? ? ?listen ? ? ? 80; > ? ? ? ?server_name ?_; > > ? ? ? ?location / { > ? ? ? ? ? ?proxy_pass_header Server; > ? ? ? ? ? ?proxy_set_header ?Host $http_host; > ? ? ? ? ? ?proxy_set_header ?X-Forwarded-For $remote_addr; > ? ? ? ? ? ?proxy_pass ? ? ? ?http://127.0.0.1:8008; > ? ? ? ?} > ? ?} > } > > > Cadaver reports the following (nautilus message isn't too informative): > > $ cadaver http://localhost/ > Could not access / (not WebDAV-enabled?): > Unknown transfer-coding in response > Connection to `localhost' closed. > dav:!> > > > I can see the request reaches pywebdav: > > localhost - - [08/Jul/2009 14:10:40] "OPTIONS / HTTP/1.0" 200 - > localhost - - [08/Jul/2009 14:10:40] "PROPFIND / HTTP/1.0" 207 - > > > Nginx access log shows: > > 127.0.0.1 - - [08/Jul/2009:14:11:20 -0700] "OPTIONS / HTTP/1.1" 200 0 "-" "cadaver/0.23.2 neon/0.28.2" > 127.0.0.1 - - [08/Jul/2009:14:11:20 -0700] "PROPFIND / HTTP/1.1" 207 593 "-" "cadaver/0.23.2 neon/0.28.2" > > > Nginx error log shows nothing. > > > Regards, > Cliff > > -- > http://www.google.com/search?q=vonage+sucks > > > From cliff at develix.com Thu Jul 9 02:27:44 2009 From: cliff at develix.com (Cliff Wells) Date: Wed, 08 Jul 2009 15:27:44 -0700 Subject: Nginx proxying WebDAV = no love In-Reply-To: References: <1247087646.3689.9.camel@portable-evil> Message-ID: <1247092064.3689.11.camel@portable-evil> I was under the impression that Nginx simply passes these along. Note that I'm proxying, not using Nginx DAV support. If this didn't work, then I'm not sure how people succeed in proxying to Apache/mod_svn, etc. Cliff On Wed, 2009-07-08 at 15:08 -0700, Michael Shadle wrote: > Does nginx even support propfind? Or options? > > I see it in the code, but I don't see anything that happens when > someone issues the verb. > > It's almost like it stubbed out for future expansion (or so it doesn't > reply with "invalid request" at least) > > On Wed, Jul 8, 2009 at 2:14 PM, Cliff Wells wrote: > > I'm experimenting with WebDAV using pywebdav: > > > > http://code.google.com/p/pywebdav/ > > > > It appears to work quite well if I connect to it > > directly (using either Nautilus and Cadaver as client). > > > > However, when I attempt to proxy to it via Nginx, > > I get messages that it's not a WebDAV-enabled share. > > > > Here's my nginx.conf: > > > > http { > > include mime.types; > > default_type application/octet-stream; > > > > server { > > listen 80; > > server_name _; > > > > location / { > > proxy_pass_header Server; > > proxy_set_header Host $http_host; > > proxy_set_header X-Forwarded-For $remote_addr; > > proxy_pass http://127.0.0.1:8008; > > } > > } > > } > > > > > > Cadaver reports the following (nautilus message isn't too informative): > > > > $ cadaver http://localhost/ > > Could not access / (not WebDAV-enabled?): > > Unknown transfer-coding in response > > Connection to `localhost' closed. > > dav:!> > > > > > > I can see the request reaches pywebdav: > > > > localhost - - [08/Jul/2009 14:10:40] "OPTIONS / HTTP/1.0" 200 - > > localhost - - [08/Jul/2009 14:10:40] "PROPFIND / HTTP/1.0" 207 - > > > > > > Nginx access log shows: > > > > 127.0.0.1 - - [08/Jul/2009:14:11:20 -0700] "OPTIONS / HTTP/1.1" 200 0 "-" "cadaver/0.23.2 neon/0.28.2" > > 127.0.0.1 - - [08/Jul/2009:14:11:20 -0700] "PROPFIND / HTTP/1.1" 207 593 "-" "cadaver/0.23.2 neon/0.28.2" > > > > > > Nginx error log shows nothing. > > > > > > Regards, > > Cliff > > > > -- > > http://www.google.com/search?q=vonage+sucks > > > > > > > -- http://www.google.com/search?q=vonage+sucks From dave at cheney.net Thu Jul 9 02:29:07 2009 From: dave at cheney.net (Dave Cheney) Date: Thu, 9 Jul 2009 08:29:07 +1000 Subject: Nginx proxying WebDAV = no love In-Reply-To: References: <1247087646.3689.9.camel@portable-evil> Message-ID: <3706FB07-0E31-412E-A68F-445D9F082F42@cheney.net> Nginx will proxy WebDAV fine, however if you are using an OS X client, nginx will send a 400 if the mac sends a Chunked Encoded request. This breaks the finder on OS X. Cheers Dave On 09/07/2009, at 8:08 AM, Michael Shadle wrote: > Does nginx even support propfind? Or options? > > I see it in the code, but I don't see anything that happens when > someone issues the verb. > > It's almost like it stubbed out for future expansion (or so it doesn't > reply with "invalid request" at least) > > On Wed, Jul 8, 2009 at 2:14 PM, Cliff Wells wrote: >> I'm experimenting with WebDAV using pywebdav: >> >> http://code.google.com/p/pywebdav/ >> >> It appears to work quite well if I connect to it >> directly (using either Nautilus and Cadaver as client). >> >> However, when I attempt to proxy to it via Nginx, >> I get messages that it's not a WebDAV-enabled share. >> >> Here's my nginx.conf: >> >> http { >> include mime.types; >> default_type application/octet-stream; >> >> server { >> listen 80; >> server_name _; >> >> location / { >> proxy_pass_header Server; >> proxy_set_header Host $http_host; >> proxy_set_header X-Forwarded-For $remote_addr; >> proxy_pass http://127.0.0.1:8008; >> } >> } >> } >> >> >> Cadaver reports the following (nautilus message isn't too >> informative): >> >> $ cadaver http://localhost/ >> Could not access / (not WebDAV-enabled?): >> Unknown transfer-coding in response >> Connection to `localhost' closed. >> dav:!> >> >> >> I can see the request reaches pywebdav: >> >> localhost - - [08/Jul/2009 14:10:40] "OPTIONS / HTTP/1.0" 200 - >> localhost - - [08/Jul/2009 14:10:40] "PROPFIND / HTTP/1.0" 207 - >> >> >> Nginx access log shows: >> >> 127.0.0.1 - - [08/Jul/2009:14:11:20 -0700] "OPTIONS / HTTP/1.1" 200 >> 0 "-" "cadaver/0.23.2 neon/0.28.2" >> 127.0.0.1 - - [08/Jul/2009:14:11:20 -0700] "PROPFIND / HTTP/1.1" >> 207 593 "-" "cadaver/0.23.2 neon/0.28.2" >> >> >> Nginx error log shows nothing. >> >> >> Regards, >> Cliff >> >> -- >> http://www.google.com/search?q=vonage+sucks >> >> >> > From nginx-forum at nginx.us Thu Jul 9 02:40:49 2009 From: nginx-forum at nginx.us (nfn) Date: Wed, 8 Jul 2009 18:40:49 -0400 Subject: Help: cache or not by cookie In-Reply-To: <90dc96ca6237b646daeeefc86ed2457a.NginxMailingList@forum.nginx.org> References: <90dc96ca6237b646daeeefc86ed2457a.NginxMailingList@forum.nginx.org> Message-ID: <65bbabb7c255ac780da15c3582911d41.NginxMailingList@forum.nginx.org> Hello, Sorry for bumping this topic but I was trying to setup a similar configuration. I want to show cached pages for a refresh of 6 min to guests (cookie user=guest) and to members (cookie user=usergroup) Am I able to read the value from the cookie? How? How do I add X-Accel-Expires: 0 to nginx? Thanks Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3173,3898#msg-3898 From mike503 at gmail.com Thu Jul 9 03:07:52 2009 From: mike503 at gmail.com (Michael Shadle) Date: Wed, 8 Jul 2009 16:07:52 -0700 Subject: Nginx proxying WebDAV = no love In-Reply-To: <1247092064.3689.11.camel@portable-evil> References: <1247087646.3689.9.camel@portable-evil> <1247092064.3689.11.camel@portable-evil> Message-ID: ah i would assume if you say proxy_pass you're literally passing *anything* over to the upstream. does nginx actually inspect it first? i was most likely only noticing the nginx DAV support, and it sets r->method but i don't see anything that happens with those actions... On Wed, Jul 8, 2009 at 3:27 PM, Cliff Wells wrote: > I was under the impression that Nginx simply passes these along. ? Note > that I'm proxying, not using Nginx DAV support. > > If this didn't work, then I'm not sure how people succeed in proxying to > Apache/mod_svn, etc. > > Cliff > > > On Wed, 2009-07-08 at 15:08 -0700, Michael Shadle wrote: >> Does nginx even support propfind? Or options? >> >> I see it in the code, but I don't see anything that happens when >> someone issues the verb. >> >> It's almost like it stubbed out for future expansion (or so it doesn't >> reply with "invalid request" at least) >> >> On Wed, Jul 8, 2009 at 2:14 PM, Cliff Wells wrote: >> > I'm experimenting with WebDAV using pywebdav: >> > >> > http://code.google.com/p/pywebdav/ >> > >> > It appears to work quite well if I connect to it >> > directly (using either Nautilus and Cadaver as client). >> > >> > However, when I attempt to proxy to it via Nginx, >> > I get messages that it's not a WebDAV-enabled share. >> > >> > Here's my nginx.conf: >> > >> > http { >> > ? ?include ? ? ? mime.types; >> > ? ?default_type ?application/octet-stream; >> > >> > ? ?server { >> > ? ? ? ?listen ? ? ? 80; >> > ? ? ? ?server_name ?_; >> > >> > ? ? ? ?location / { >> > ? ? ? ? ? ?proxy_pass_header Server; >> > ? ? ? ? ? ?proxy_set_header ?Host $http_host; >> > ? ? ? ? ? ?proxy_set_header ?X-Forwarded-For $remote_addr; >> > ? ? ? ? ? ?proxy_pass ? ? ? ?http://127.0.0.1:8008; >> > ? ? ? ?} >> > ? ?} >> > } >> > >> > >> > Cadaver reports the following (nautilus message isn't too informative): >> > >> > $ cadaver http://localhost/ >> > Could not access / (not WebDAV-enabled?): >> > Unknown transfer-coding in response >> > Connection to `localhost' closed. >> > dav:!> >> > >> > >> > I can see the request reaches pywebdav: >> > >> > localhost - - [08/Jul/2009 14:10:40] "OPTIONS / HTTP/1.0" 200 - >> > localhost - - [08/Jul/2009 14:10:40] "PROPFIND / HTTP/1.0" 207 - >> > >> > >> > Nginx access log shows: >> > >> > 127.0.0.1 - - [08/Jul/2009:14:11:20 -0700] "OPTIONS / HTTP/1.1" 200 0 "-" "cadaver/0.23.2 neon/0.28.2" >> > 127.0.0.1 - - [08/Jul/2009:14:11:20 -0700] "PROPFIND / HTTP/1.1" 207 593 "-" "cadaver/0.23.2 neon/0.28.2" >> > >> > >> > Nginx error log shows nothing. >> > >> > >> > Regards, >> > Cliff >> > >> > -- >> > http://www.google.com/search?q=vonage+sucks >> > >> > >> > >> > -- > http://www.google.com/search?q=vonage+sucks > > > From chanceycn at gmail.com Thu Jul 9 05:48:54 2009 From: chanceycn at gmail.com (Chancey) Date: Thu, 9 Jul 2009 09:48:54 +0800 Subject: How to write this rewrite? Message-ID: <200907090948443126138@gmail.com> Hi, guys! This is Apache's format, how to write in nginx? Thanks a lot. RewriteEngine On RewriteCond %{REQUEST_FILENAME} !-f RewriteCond %{REQUEST_FILENAME} !-d RewriteRule !\.(js|ico|gif|jpg|png|css|cgi|html|htm|cert|swf)$ index.php php_flag always_populate_raw_post_data On 2009-07-09 Chancey -------------- next part -------------- An HTML attachment was scrubbed... URL: From is at rambler-co.ru Thu Jul 9 11:02:44 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Thu, 9 Jul 2009 11:02:44 +0400 Subject: How to write this rewrite? In-Reply-To: <200907090948443126138@gmail.com> References: <200907090948443126138@gmail.com> Message-ID: <20090709070244.GB66728@rambler-co.ru> On Thu, Jul 09, 2009 at 09:48:54AM +0800, Chancey wrote: > Hi, guys! > > This is Apache's format, how to write in nginx? Thanks a lot. > > > RewriteEngine On > RewriteCond %{REQUEST_FILENAME} !-f > RewriteCond %{REQUEST_FILENAME} !-d > RewriteRule !\.(js|ico|gif|jpg|png|css|cgi|html|htm|cert|swf)$ index.php > php_flag always_populate_raw_post_data On > Probably something like this: location / { try_files $uri $uri/ /index.php; fastcgi_pass ... } location ~ \.(js|ico|gif|jpg|png|css|cgi|html|htm|cert|swf)$ { } -- Igor Sysoev http://sysoev.ru/en/ From is at rambler-co.ru Thu Jul 9 11:00:09 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Thu, 9 Jul 2009 11:00:09 +0400 Subject: Nginx proxying WebDAV = no love In-Reply-To: <1247087646.3689.9.camel@portable-evil> References: <1247087646.3689.9.camel@portable-evil> Message-ID: <20090709070009.GA66728@rambler-co.ru> On Wed, Jul 08, 2009 at 02:14:06PM -0700, Cliff Wells wrote: > I'm experimenting with WebDAV using pywebdav: > > http://code.google.com/p/pywebdav/ > > It appears to work quite well if I connect to it > directly (using either Nautilus and Cadaver as client). > > However, when I attempt to proxy to it via Nginx, > I get messages that it's not a WebDAV-enabled share. > > Here's my nginx.conf: > > http { > include mime.types; > default_type application/octet-stream; > > server { > listen 80; > server_name _; > > location / { > proxy_pass_header Server; > proxy_set_header Host $http_host; > proxy_set_header X-Forwarded-For $remote_addr; > proxy_pass http://127.0.0.1:8008; > } > } > } > > > Cadaver reports the following (nautilus message isn't too informative): > > $ cadaver http://localhost/ > Could not access / (not WebDAV-enabled?): > Unknown transfer-coding in response > Connection to `localhost' closed. > dav:!> > > > I can see the request reaches pywebdav: > > localhost - - [08/Jul/2009 14:10:40] "OPTIONS / HTTP/1.0" 200 - > localhost - - [08/Jul/2009 14:10:40] "PROPFIND / HTTP/1.0" 207 - > > > Nginx access log shows: > > 127.0.0.1 - - [08/Jul/2009:14:11:20 -0700] "OPTIONS / HTTP/1.1" 200 0 "-" "cadaver/0.23.2 neon/0.28.2" > 127.0.0.1 - - [08/Jul/2009:14:11:20 -0700] "PROPFIND / HTTP/1.1" 207 593 "-" "cadaver/0.23.2 neon/0.28.2" > > > Nginx error log shows nothing. Coud you create debug log of the requests ? -- Igor Sysoev http://sysoev.ru/en/ From is at rambler-co.ru Thu Jul 9 11:05:16 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Thu, 9 Jul 2009 11:05:16 +0400 Subject: Help: cache or not by cookie In-Reply-To: <65bbabb7c255ac780da15c3582911d41.NginxMailingList@forum.nginx.org> References: <90dc96ca6237b646daeeefc86ed2457a.NginxMailingList@forum.nginx.org> <65bbabb7c255ac780da15c3582911d41.NginxMailingList@forum.nginx.org> Message-ID: <20090709070516.GC66728@rambler-co.ru> On Wed, Jul 08, 2009 at 06:40:49PM -0400, nfn wrote: > Hello, > > Sorry for bumping this topic but I was trying to setup a similar configuration. > I want to show cached pages for a refresh of 6 min to guests (cookie user=guest) and to members (cookie user=usergroup) proxy_cache_key "http://cacheserver$request_uri $cookie_user"; proxy_cache_valid 6m; > Am I able to read the value from the cookie? How? > How do I add X-Accel-Expires: 0 to nginx? You should add this header in backend. -- Igor Sysoev http://sysoev.ru/en/ From chanceycn at gmail.com Thu Jul 9 11:34:32 2009 From: chanceycn at gmail.com (=?utf-8?B?Q2hhbmNleQ==?=) Date: Thu, 9 Jul 2009 15:34:32 +0800 Subject: =?utf-8?B?UmU6IFJlOiBIb3cgdG8gd3JpdGUgdGhpcyByZXdyaXRlPw==?= References: <200907090948443126138@gmail.com>, <20090709070244.GB66728@rambler-co.ru> Message-ID: <200907091534266715615@gmail.com> Thanks, i'll try it 2009-07-09 Chancey ???? Igor Sysoev ????? 2009-07-09 15:10:56 ???? nginx ??? ??? Re: How to write this rewrite? On Thu, Jul 09, 2009 at 09:48:54AM +0800, Chancey wrote: > Hi, guys! > > This is Apache's format, how to write in nginx? Thanks a lot. > > > RewriteEngine On > RewriteCond %{REQUEST_FILENAME} !-f > RewriteCond %{REQUEST_FILENAME} !-d > RewriteRule !\.(js|ico|gif|jpg|png|css|cgi|html|htm|cert|swf)$ index.php > php_flag always_populate_raw_post_data On > Probably something like this: location / { try_files $uri $uri/ /index.php; fastcgi_pass ... } location ~ \.(js|ico|gif|jpg|png|css|cgi|html|htm|cert|swf)$ { } -- Igor Sysoev http://sysoev.ru/en/ -------------- next part -------------- An HTML attachment was scrubbed... URL: From miles at tinyapps.org Thu Jul 9 11:42:05 2009 From: miles at tinyapps.org (TinyApps.Org) Date: Wed, 8 Jul 2009 21:42:05 -1000 Subject: Password-protected dirs working but reporting unexpected errors Message-ID: I sent this out a week ago and received no reply, so I am trying a second and final time to see if anyone can assist: Password protected directories are working fine, but when a legitimate user logs in, the following appears in the error log: [error] 3876#0: *12 no user/password was provided for basic authentication despite the fact that the username and password *were* supplied, and access was granted. A bad username/password produces the expected error: [error] 3876#0: *12 user "whatever" was not found in "/etc/nginx/htpass" Has anyone else seen this or have any idea why such an error would be triggered? From is at rambler-co.ru Thu Jul 9 11:58:29 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Thu, 9 Jul 2009 11:58:29 +0400 Subject: Password-protected dirs working but reporting unexpected errors In-Reply-To: References: Message-ID: <20090709075829.GE66728@rambler-co.ru> On Wed, Jul 08, 2009 at 09:42:05PM -1000, TinyApps.Org wrote: > I sent this out a week ago and received no reply, > so I am trying a second and final time to see if > anyone can assist: > > Password protected directories are working fine, > but when a legitimate user logs in, the following > appears in the error log: > > [error] 3876#0: *12 no user/password was provided for basic > authentication > > despite the fact that the username and password > *were* supplied, and access was granted. The HTTP Basic authentication works as following: *) A browser requests a page without user/password. *) A server response with 401 page, sending realm as well. At this stage the 401 code appears in access_log and the message "no user/password ..." appears in error_log. *) The browser shows a realm/login/password prompt. *) If a user will press cancel, then the browser will show the received 401 page. *) If the user enters login/password, then the browser repeats the request with login/password. Then until you will exit the browser, it will send these login/password with all requests in protected hierarchy. > A bad username/password produces the expected error: > > [error] 3876#0: *12 user "whatever" was not found in "/etc/nginx/htpass" > > Has anyone else seen this or have any idea why such > an error would be triggered? -- Igor Sysoev http://sysoev.ru/en/ From dave at cheney.net Thu Jul 9 12:30:29 2009 From: dave at cheney.net (Dave Cheney) Date: Thu, 9 Jul 2009 18:30:29 +1000 Subject: Nginx proxying WebDAV = no love In-Reply-To: <1247087646.3689.9.camel@portable-evil> References: <1247087646.3689.9.camel@portable-evil> Message-ID: Get out ngrep and make sure that pywebdav is honoring the Connection: close request proxied via nginx. I have found that nginx waits for the socket to close before forwarding the request. This is correct behaviour for a HTTP/1.0 proxy, but most servers are used to HTTP/1.1 clients who are less correct in their handling of HTTP/1.0. Cheers Dave On 09/07/2009, at 7:14 AM, Cliff Wells wrote: > I'm experimenting with WebDAV using pywebdav: > > http://code.google.com/p/pywebdav/ > > It appears to work quite well if I connect to it > directly (using either Nautilus and Cadaver as client). > > However, when I attempt to proxy to it via Nginx, > I get messages that it's not a WebDAV-enabled share. > > Here's my nginx.conf: > > http { > include mime.types; > default_type application/octet-stream; > > server { > listen 80; > server_name _; > > location / { > proxy_pass_header Server; > proxy_set_header Host $http_host; > proxy_set_header X-Forwarded-For $remote_addr; > proxy_pass http://127.0.0.1:8008; > } > } > } > > > Cadaver reports the following (nautilus message isn't too > informative): > > $ cadaver http://localhost/ > Could not access / (not WebDAV-enabled?): > Unknown transfer-coding in response > Connection to `localhost' closed. > dav:!> > > > I can see the request reaches pywebdav: > > localhost - - [08/Jul/2009 14:10:40] "OPTIONS / HTTP/1.0" 200 - > localhost - - [08/Jul/2009 14:10:40] "PROPFIND / HTTP/1.0" 207 - > > > Nginx access log shows: > > 127.0.0.1 - - [08/Jul/2009:14:11:20 -0700] "OPTIONS / HTTP/1.1" 200 > 0 "-" "cadaver/0.23.2 neon/0.28.2" > 127.0.0.1 - - [08/Jul/2009:14:11:20 -0700] "PROPFIND / HTTP/1.1" 207 > 593 "-" "cadaver/0.23.2 neon/0.28.2" > > > Nginx error log shows nothing. > > > Regards, > Cliff > > -- > http://www.google.com/search?q=vonage+sucks > > From jg at internetx.de Thu Jul 9 12:35:45 2009 From: jg at internetx.de (Juergen Gotteswinter) Date: Thu, 09 Jul 2009 10:35:45 +0200 Subject: nginx rewrite question Message-ID: <4A55ABE1.7080704@internetx.de> Hello, perhaps someone here could help me with the following problem. i got a domain justfornow.com which got a wildcard dns entry, and i need to redirect this domain to https. but i have to keep the first part of the domain, for example: user calls http://try.justfornow.com nginx rewrites it to https://try.justfornow.com user calls http://anything.justfornow.com nginx keeps the first part and rewrites it to https://anything.justfornow.com thanks :) From is at rambler-co.ru Thu Jul 9 12:40:29 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Thu, 9 Jul 2009 12:40:29 +0400 Subject: nginx rewrite question In-Reply-To: <4A55ABE1.7080704@internetx.de> References: <4A55ABE1.7080704@internetx.de> Message-ID: <20090709084029.GJ66728@rambler-co.ru> On Thu, Jul 09, 2009 at 10:35:45AM +0200, Juergen Gotteswinter wrote: > Hello, > > perhaps someone here could help me with the following problem. i got a > domain justfornow.com which got a wildcard dns entry, and i need to > redirect this domain to https. but i have to keep the first part of the > domain, for example: > > user calls http://try.justfornow.com nginx rewrites it to > https://try.justfornow.com > user calls http://anything.justfornow.com nginx keeps the first part and > rewrites it to https://anything.justfornow.com server { listen 80; server_name *.justfornow.com; rewrite ^ https://$host$request_uri? permanent; } -- Igor Sysoev http://sysoev.ru/en/ From nginx-forum at nginx.us Thu Jul 9 14:05:26 2009 From: nginx-forum at nginx.us (nfn) Date: Thu, 9 Jul 2009 06:05:26 -0400 Subject: Help: cache or not by cookie In-Reply-To: <90dc96ca6237b646daeeefc86ed2457a.NginxMailingList@forum.nginx.org> References: <90dc96ca6237b646daeeefc86ed2457a.NginxMailingList@forum.nginx.org> Message-ID: <9edb91412795487e4f2f00c6ab99e755.NginxMailingList@forum.nginx.org> Thanks Igor, but how do I validate if the user is guest. I only want to show the cached version to guests. Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3173,3923#msg-3923 From alouche07 at gmail.com Thu Jul 9 16:24:21 2009 From: alouche07 at gmail.com (Ali Abbas) Date: Thu, 9 Jul 2009 14:24:21 +0200 Subject: [emerg]: unknown directive "upstream" Message-ID: Hello, Isn't the http://wiki.nginx.org/NginxHttpUpstreamModule supposed to be activated by default? Thanks, -------------- next part -------------- An HTML attachment was scrubbed... URL: From mdounin at mdounin.ru Thu Jul 9 17:05:52 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Thu, 9 Jul 2009 17:05:52 +0400 Subject: [emerg]: unknown directive "upstream" In-Reply-To: References: Message-ID: <20090709130552.GL54083@mdounin.ru> Hello! On Thu, Jul 09, 2009 at 02:24:21PM +0200, Ali Abbas wrote: > Hello, > > Isn't the http://wiki.nginx.org/NginxHttpUpstreamModule supposed to be > activated by default? It's compiled in by default. Check your config syntax - looks like you used upstream{} at wrong level (not within http block?). Maxim Dounin From alouche07 at gmail.com Thu Jul 9 17:36:08 2009 From: alouche07 at gmail.com (Ali Abbas) Date: Thu, 9 Jul 2009 15:36:08 +0200 Subject: [emerg]: unknown directive "upstream" In-Reply-To: <20090709130552.GL54083@mdounin.ru> References: <20090709130552.GL54083@mdounin.ru> Message-ID: Hi, > > Hello, > > > > Isn't the http://wiki.nginx.org/NginxHttpUpstreamModule supposed to be > > activated by default? > > It's compiled in by default. Check your config syntax - looks > like you used upstream{} at wrong level (not within http block?). Yeah that was it. Thx, -------------- next part -------------- An HTML attachment was scrubbed... URL: From miles at tinyapps.org Thu Jul 9 18:25:14 2009 From: miles at tinyapps.org (TinyApps.Org) Date: Thu, 9 Jul 2009 04:25:14 -1000 Subject: Password-protected dirs working but reporting unexpected errors In-Reply-To: <20090709075829.GE66728@rambler-co.ru> References: <20090709075829.GE66728@rambler-co.ru> Message-ID: Thanks so much, Igor! I'd like to send you a donation, but do not see a button on http://sysoev.ru/en/ or http://nginx.net/ . Do the donations to http://wiki.nginx.org/Main really go to you? Just want to make sure. Yours Gratefully, Miles On Jul 8, 2009, at 9:58 PM, Igor Sysoev wrote: > On Wed, Jul 08, 2009 at 09:42:05PM -1000, TinyApps.Org wrote: > >> I sent this out a week ago and received no reply, >> so I am trying a second and final time to see if >> anyone can assist: >> >> Password protected directories are working fine, >> but when a legitimate user logs in, the following >> appears in the error log: >> >> [error] 3876#0: *12 no user/password was provided for basic >> authentication >> >> despite the fact that the username and password >> *were* supplied, and access was granted. > > The HTTP Basic authentication works as following: > *) A browser requests a page without user/password. > *) A server response with 401 page, sending realm as well. > At this stage the 401 code appears in access_log and the message > "no user/password ..." appears in error_log. > *) The browser shows a realm/login/password prompt. > *) If a user will press cancel, then the browser will show the > received > 401 page. > *) If the user enters login/password, then the browser repeats the > request > with login/password. > > Then until you will exit the browser, it will send these login/ > password > with all requests in protected hierarchy. > >> A bad username/password produces the expected error: >> >> [error] 3876#0: *12 user "whatever" was not found in "/etc/nginx/ >> htpass" >> >> Has anyone else seen this or have any idea why such >> an error would be triggered? > > > -- > Igor Sysoev > http://sysoev.ru/en/ > From is at rambler-co.ru Thu Jul 9 18:26:36 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Thu, 9 Jul 2009 18:26:36 +0400 Subject: Help: cache or not by cookie In-Reply-To: <9edb91412795487e4f2f00c6ab99e755.NginxMailingList@forum.nginx.org> References: <90dc96ca6237b646daeeefc86ed2457a.NginxMailingList@forum.nginx.org> <9edb91412795487e4f2f00c6ab99e755.NginxMailingList@forum.nginx.org> Message-ID: <20090709142636.GE72073@rambler-co.ru> On Thu, Jul 09, 2009 at 06:05:26AM -0400, nfn wrote: > Thanks Igor, but how do I validate if the user is guest. I only want to show the cached version to guests. You need to use proxy_cache_key "http://cacheserver$request_uri $cookie_user"; proxy_cache_valid 6m; Then in your backend you should test the "user" cookie. If it is not equal to "guest", then you should add the "X-Accel-Redirect: 0" header in a response. In this case nginx will not cache the response. Otherwise, if cookie is equal to "guest" you do not need to set the header or may set "X-Accel-Redirect: 360" to say nginx to cache the response to 6 min. -- Igor Sysoev http://sysoev.ru/en/ From is at rambler-co.ru Thu Jul 9 18:31:39 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Thu, 9 Jul 2009 18:31:39 +0400 Subject: vburshteyn In-Reply-To: <2c449df47c612bdeaa42d965f6c610af.NginxMailingList@forum.nginx.org> References: <2c449df47c612bdeaa42d965f6c610af.NginxMailingList@forum.nginx.org> Message-ID: <20090709143139.GF72073@rambler-co.ru> On Wed, Jul 08, 2009 at 09:18:24AM -0400, vburshteyn wrote: > Hi folks. back again for more info :) > > I need to use the rewrite module to rewrite redirects. > > for example > > http://www.mysite.com/9-to-5/group_show/569889 needs to be http://groups.bqa.mysite.com/9-to-5/group_show/569889 > > any idea how i would go about doing this? If you need just to redirect to another host, then: server { server_name www.mysite.com; rewrite ^ http://groups.bqa.mysite.com$request_uri?; } -- Igor Sysoev http://sysoev.ru/en/ From kgorlo at gmail.com Thu Jul 9 18:47:43 2009 From: kgorlo at gmail.com (Kamil Gorlo) Date: Thu, 9 Jul 2009 16:47:43 +0200 Subject: Upstream (fair) with dead servers Message-ID: <1c14c1730907090747w4294ecffm3d1fc3d3c9010b0f@mail.gmail.com> Hi, I have (possibly quite rare) situation where I have one upstream with two servers and I know that in any particular moment only one of those servers is running: upstream blah { server server1; // when working, server2 is dead server server2; // when working, server1 is dead } So, what I need is that Nginx will choose working backend (in this upstream), remembers it and choose it for every request till some failure on this server occurs. Then it tries other server in upstream, till it finds some working - and situation repeats. Is this possible? Maybe upstream_fair module with option no_rr will work for me (I am not sure)? Cheers, -- Kamil Gorlo From is at rambler-co.ru Thu Jul 9 18:48:35 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Thu, 9 Jul 2009 18:48:35 +0400 Subject: Password-protected dirs working but reporting unexpected errors In-Reply-To: References: <20090709075829.GE66728@rambler-co.ru> Message-ID: <20090709144835.GG72073@rambler-co.ru> On Thu, Jul 09, 2009 at 04:25:14AM -1000, TinyApps.Org wrote: > Thanks so much, Igor! > I'd like to send you a donation, but do not see > a button on http://sysoev.ru/en/ or http://nginx.net/ . > Do the donations to http://wiki.nginx.org/Main really > go to you? Just want to make sure. Yes, http://sysoev.ru/en/donation.html > Yours Gratefully, > > Miles > > > On Jul 8, 2009, at 9:58 PM, Igor Sysoev wrote: > > >On Wed, Jul 08, 2009 at 09:42:05PM -1000, TinyApps.Org wrote: > > > >>I sent this out a week ago and received no reply, > >>so I am trying a second and final time to see if > >>anyone can assist: > >> > >>Password protected directories are working fine, > >>but when a legitimate user logs in, the following > >>appears in the error log: > >> > >>[error] 3876#0: *12 no user/password was provided for basic > >>authentication > >> > >>despite the fact that the username and password > >>*were* supplied, and access was granted. > > > >The HTTP Basic authentication works as following: > >*) A browser requests a page without user/password. > >*) A server response with 401 page, sending realm as well. > > At this stage the 401 code appears in access_log and the message > > "no user/password ..." appears in error_log. > >*) The browser shows a realm/login/password prompt. > >*) If a user will press cancel, then the browser will show the > >received > > 401 page. > >*) If the user enters login/password, then the browser repeats the > >request > > with login/password. > > > >Then until you will exit the browser, it will send these login/ > >password > >with all requests in protected hierarchy. > > > >>A bad username/password produces the expected error: > >> > >>[error] 3876#0: *12 user "whatever" was not found in "/etc/nginx/ > >>htpass" > >> > >>Has anyone else seen this or have any idea why such > >>an error would be triggered? > > > > > >-- > >Igor Sysoev > >http://sysoev.ru/en/ > > > -- ????? ?????? http://sysoev.ru From is at rambler-co.ru Thu Jul 9 18:57:01 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Thu, 9 Jul 2009 18:57:01 +0400 Subject: Upstream (fair) with dead servers In-Reply-To: <1c14c1730907090747w4294ecffm3d1fc3d3c9010b0f@mail.gmail.com> References: <1c14c1730907090747w4294ecffm3d1fc3d3c9010b0f@mail.gmail.com> Message-ID: <20090709145701.GH72073@rambler-co.ru> On Thu, Jul 09, 2009 at 04:47:43PM +0200, Kamil Gorlo wrote: > Hi, > > I have (possibly quite rare) situation where I have one upstream with > two servers and I know that in any particular moment only one of those > servers is running: > > upstream blah { > server server1; // when working, server2 is dead > server server2; // when working, server1 is dead > } > > So, what I need is that Nginx will choose working backend (in this > upstream), remembers it and choose it for every request till some > failure on this server occurs. Then it tries other server in upstream, > till it finds some working - and situation repeats. Is this possible? No, nginx will send request on both servers in round robin mode. If one of them will fails, then nginx will resend the failed request to another server and mark the server as dead for fail_timeout time. > Maybe upstream_fair module with option no_rr will work for me (I am not sure)? > > Cheers, > -- > Kamil Gorlo -- Igor Sysoev http://sysoev.ru/en/ From kgorlo at gmail.com Thu Jul 9 19:09:56 2009 From: kgorlo at gmail.com (Kamil Gorlo) Date: Thu, 9 Jul 2009 17:09:56 +0200 Subject: Upstream (fair) with dead servers In-Reply-To: <20090709145701.GH72073@rambler-co.ru> References: <1c14c1730907090747w4294ecffm3d1fc3d3c9010b0f@mail.gmail.com> <20090709145701.GH72073@rambler-co.ru> Message-ID: <1c14c1730907090809r66e6c2abm94a5f2a74bc7079a@mail.gmail.com> 2009/7/9 Igor Sysoev : > On Thu, Jul 09, 2009 at 04:47:43PM +0200, Kamil Gorlo wrote: > >> Hi, >> >> I have (possibly quite rare) situation where I have one upstream with >> two servers and I know that in any particular moment only one of those >> servers is running: >> >> upstream blah { >> ? ? server server1; // when working, server2 is dead >> ? ? server server2; // when working, server1 is dead >> } >> >> So, what I need is that Nginx will choose working backend (in this >> upstream), remembers it and choose it for every request till some >> failure on this server occurs. Then it tries other server in upstream, >> till it finds some working - and situation repeats. Is this possible? > > No, nginx will send request on both servers in round robin mode. > If one of them will fails, then nginx will resend the failed request > to another server and mark the server as dead for fail_timeout time. Hmm, what about situation when all servers are marked as dead? Nginx will wait till fail_timeout expires for some server and in meantime return error to client? Or will try dead servers? Also, is it hard to write such a "upstream-chooser" which will work as I described? >> Maybe upstream_fair module with option no_rr will work for me (I am not sure)? >> >> Cheers, >> -- >> Kamil Gorlo > > > -- > Igor Sysoev > http://sysoev.ru/en/ > > From is at rambler-co.ru Thu Jul 9 19:14:54 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Thu, 9 Jul 2009 19:14:54 +0400 Subject: Upstream (fair) with dead servers In-Reply-To: <1c14c1730907090809r66e6c2abm94a5f2a74bc7079a@mail.gmail.com> References: <1c14c1730907090747w4294ecffm3d1fc3d3c9010b0f@mail.gmail.com> <20090709145701.GH72073@rambler-co.ru> <1c14c1730907090809r66e6c2abm94a5f2a74bc7079a@mail.gmail.com> Message-ID: <20090709151454.GI72073@rambler-co.ru> On Thu, Jul 09, 2009 at 05:09:56PM +0200, Kamil Gorlo wrote: > 2009/7/9 Igor Sysoev : > > On Thu, Jul 09, 2009 at 04:47:43PM +0200, Kamil Gorlo wrote: > > > >> Hi, > >> > >> I have (possibly quite rare) situation where I have one upstream with > >> two servers and I know that in any particular moment only one of those > >> servers is running: > >> > >> upstream blah { > >> ? ? server server1; // when working, server2 is dead > >> ? ? server server2; // when working, server1 is dead > >> } > >> > >> So, what I need is that Nginx will choose working backend (in this > >> upstream), remembers it and choose it for every request till some > >> failure on this server occurs. Then it tries other server in upstream, > >> till it finds some working - and situation repeats. Is this possible? > > > > No, nginx will send request on both servers in round robin mode. > > If one of them will fails, then nginx will resend the failed request > > to another server and mark the server as dead for fail_timeout time. > > Hmm, what about situation when all servers are marked as dead? Nginx > will wait till fail_timeout expires for some server and in meantime > return error to client? Or will try dead servers? In this case nginx marks all servers as live and try again. > Also, is it hard to write such a "upstream-chooser" which will work as > I described? I'm going to rewrite upstream modules in 0.8.x (add fair, etc.) and probably add this functionality. But why do you want to send request to one only server leaving the other as spare one ? > >> Maybe upstream_fair module with option no_rr will work for me (I am not sure)? > >> > >> Cheers, > >> -- > >> Kamil Gorlo -- Igor Sysoev http://sysoev.rua/en/ From miles at tinyapps.org Thu Jul 9 19:20:54 2009 From: miles at tinyapps.org (TinyApps.Org) Date: Thu, 9 Jul 2009 05:20:54 -1000 Subject: Password-protected dirs working but reporting unexpected errors In-Reply-To: <20090709144835.GG72073@rambler-co.ru> References: <20090709075829.GE66728@rambler-co.ru> <20090709144835.GG72073@rambler-co.ru> Message-ID: >> I'd like to send you a donation, but do not see >> a button on http://sysoev.ru/en/ or http://nginx.net/ . >> Do the donations to http://wiki.nginx.org/Main really >> go to you? Just want to make sure. > Yes, http://sysoev.ru/en/donation.html Thanks - done! From kgorlo at gmail.com Thu Jul 9 19:46:20 2009 From: kgorlo at gmail.com (Kamil Gorlo) Date: Thu, 9 Jul 2009 17:46:20 +0200 Subject: Upstream (fair) with dead servers In-Reply-To: <20090709151454.GI72073@rambler-co.ru> References: <1c14c1730907090747w4294ecffm3d1fc3d3c9010b0f@mail.gmail.com> <20090709145701.GH72073@rambler-co.ru> <1c14c1730907090809r66e6c2abm94a5f2a74bc7079a@mail.gmail.com> <20090709151454.GI72073@rambler-co.ru> Message-ID: <1c14c1730907090846u331ec5d3v144fbe65351d9ae@mail.gmail.com> 2009/7/9 Igor Sysoev : > On Thu, Jul 09, 2009 at 05:09:56PM +0200, Kamil Gorlo wrote: > >> 2009/7/9 Igor Sysoev : >> > On Thu, Jul 09, 2009 at 04:47:43PM +0200, Kamil Gorlo wrote: >> > >> >> Hi, >> >> >> >> I have (possibly quite rare) situation where I have one upstream with >> >> two servers and I know that in any particular moment only one of those >> >> servers is running: >> >> >> >> upstream blah { >> >> ? ? server server1; // when working, server2 is dead >> >> ? ? server server2; // when working, server1 is dead >> >> } >> >> >> >> So, what I need is that Nginx will choose working backend (in this >> >> upstream), remembers it and choose it for every request till some >> >> failure on this server occurs. Then it tries other server in upstream, >> >> till it finds some working - and situation repeats. Is this possible? >> > >> > No, nginx will send request on both servers in round robin mode. >> > If one of them will fails, then nginx will resend the failed request >> > to another server and mark the server as dead for fail_timeout time. >> >> Hmm, what about situation when all servers are marked as dead? Nginx >> will wait till fail_timeout expires for some server and in meantime >> return error to client? Or will try dead servers? > > In this case nginx marks all servers as live and try again. > >> Also, is it hard to write such a "upstream-chooser" which will work as >> I described? > > I'm going to rewrite upstream modules in 0.8.x (add fair, etc.) and > probably add this functionality. But why do you want to send request > to one only server leaving the other as spare one ? Those servers are not HTTP servers, but memcached-like servers (Tokyo Tyrant). In my configuration Tokyo Tyrant is in master-master taste - which leads to problems with replication when clients are writing to more than one master (but this is master-master not master-slave because I want that, in case of failure, the other working server will cope with writes). So to prevent problems I have simple (and maybe ugly;) scripts which blocks interfaces at non-current-master machine. >> >> Maybe upstream_fair module with option no_rr will work for me (I am not sure)? >> >> >> >> Cheers, >> >> -- >> >> Kamil Gorlo > > > -- > Igor Sysoev > http://sysoev.rua/en/ > > From nginx at tomvalentine.net Thu Jul 9 22:08:46 2009 From: nginx at tomvalentine.net (nginx at tomvalentine.net) Date: Thu, 09 Jul 2009 19:08:46 +0100 Subject: Auth_basic 500 error Message-ID: <4A56322E.6020703@tomvalentine.net> Hello, I can not seem to get auth basic to work, my configuration works fine with everything else, but if I enable authorisation, I get a 500 Internal Server Error I think the key error is the Readfile() error, but I have no idea what it means or how to fix it. From nginx.conf location / { index index.html index.htm index.php; auth_basic private; auth_basic_user_file htpasswd; } If anyone can help, thanks! By the way, I am using nginx/0.7.61 on Windows Vista, and the error happens in both Firefox/Opera Tom Attached is part of my log file -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: 500_error.log URL: From mike503 at gmail.com Thu Jul 9 22:32:01 2009 From: mike503 at gmail.com (Michael Shadle) Date: Thu, 9 Jul 2009 11:32:01 -0700 Subject: Auth_basic 500 error In-Reply-To: <4A56322E.6020703@tomvalentine.net> References: <4A56322E.6020703@tomvalentine.net> Message-ID: try the full path to the htpasswd file. it's probably trying to read from the nginx config path. On Thu, Jul 9, 2009 at 11:08 AM, wrote: > Hello, > > I can not seem to get auth basic to work, my configuration works fine > with everything else, but if I enable authorisation, I get a 500 > Internal Server Error > > I think the key error is the Readfile() error, but I have no idea what > it means or how to fix it. > > ?From nginx.conf > > location / { > ? ? ? ?index ?index.html index.htm index.php; > ? ? ? ?auth_basic private; > ? ? ? ?auth_basic_user_file htpasswd; > } > > If anyone can help, thanks! > By the way, I am using nginx/0.7.61 on Windows Vista, and the error > happens in both Firefox/Opera > > Tom > > Attached is part of my log file > > > > > 2009/07/09 17:01:45 [debug] 1348#504: *3 http request line: "GET / HTTP/1.1" > 2009/07/09 17:01:45 [debug] 1348#504: *3 http uri: "/" > 2009/07/09 17:01:45 [debug] 1348#504: *3 http args: "" > 2009/07/09 17:01:45 [debug] 1348#504: *3 http exten: "" > 2009/07/09 17:01:45 [debug] 1348#504: *3 http process request header line > 2009/07/09 17:01:45 [debug] 1348#504: *3 http header: "Host: localhost:88" > 2009/07/09 17:01:45 [debug] 1348#504: *3 http header: "User-Agent: > Mozilla/5.0 (Windows; U; Windows NT 6.0; en-GB; rv:1.9.1) Gecko/20090624 > Firefox/3.5 (.NET CLR 3.5.30729)" > 2009/07/09 17:01:45 [debug] 1348#504: *3 http header: "Accept: > text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8" > 2009/07/09 17:01:45 [debug] 1348#504: *3 http header: "Accept-Language: > en-gb,en;q=0.5" > 2009/07/09 17:01:45 [debug] 1348#504: *3 http header: "Accept-Encoding: > gzip,deflate" > 2009/07/09 17:01:45 [debug] 1348#504: *3 http header: "Accept-Charset: > ISO-8859-1,utf-8;q=0.7,*;q=0.7" > 2009/07/09 17:01:45 [debug] 1348#504: *3 http header: "Keep-Alive: 300" > 2009/07/09 17:01:45 [debug] 1348#504: *3 http header: "Connection: > keep-alive" > 2009/07/09 17:01:45 [debug] 1348#504: *3 http header: "Authorization: Basic > dG9tOmhramhmc2RsZmpo" > 2009/07/09 17:01:45 [debug] 1348#504: *3 http header: "Pragma: no-cache" > 2009/07/09 17:01:45 [debug] 1348#504: *3 http header: "Cache-Control: > no-cache" > 2009/07/09 17:01:45 [debug] 1348#504: *3 http header done > 2009/07/09 17:01:45 [debug] 1348#504: *3 event timer del: 320: 1614849850 > 2009/07/09 17:01:45 [debug] 1348#504: *3 generic phase: 0 > 2009/07/09 17:01:45 [debug] 1348#504: *3 add cleanup: 018BD910 > 2009/07/09 17:01:45 [debug] 1348#504: *3 generic phase: 1 > 2009/07/09 17:01:45 [debug] 1348#504: *3 test location: "/" > 2009/07/09 17:01:45 [debug] 1348#504: *3 using configuration "/" > 2009/07/09 17:01:45 [debug] 1348#504: *3 http cl:-1 max:1048576 > 2009/07/09 17:01:45 [debug] 1348#504: *3 generic phase: 3 > 2009/07/09 17:01:45 [debug] 1348#504: *3 post rewrite phase: 4 > 2009/07/09 17:01:45 [debug] 1348#504: *3 generic phase: 5 > 2009/07/09 17:01:45 [debug] 1348#504: *3 add cleanup: 018BD92C > 2009/07/09 17:01:45 [debug] 1348#504: *3 generic phase: 6 > 2009/07/09 17:01:45 [debug] 1348#504: *3 generic phase: 7 > 2009/07/09 17:01:45 [debug] 1348#504: *3 access phase: 8 > 2009/07/09 17:01:45 [debug] 1348#504: *3 access: 0100007F FFFFFFFF 0100007F > 2009/07/09 17:01:45 [debug] 1348#504: *3 access phase: 9 > 2009/07/09 17:01:45 [error] 1348#504: *3 ReadFile() failed (38: Reached the > end of the file), client: 127.0.0.1, server: localhost, request: "GET / > HTTP/1.1", host: "localhost:88" > 2009/07/09 17:01:45 [debug] 1348#504: *3 http finalize request: 500, "/?" 1 > 2009/07/09 17:01:45 [debug] 1348#504: *3 http special response: 500, "/?" > 2009/07/09 17:01:45 [debug] 1348#504: *3 http set discard body > 2009/07/09 17:01:45 [debug] 1348#504: *3 HTTP/1.1 500 Internal Server Error > Server: nginx/0.7.61 > Date: Thu, 09 Jul 2009 16:01:45 GMT > Content-Type: text/html > Content-Length: 193 > Connection: close > > From nginx-forum at nginx.us Thu Jul 9 23:33:14 2009 From: nginx-forum at nginx.us (vburshteyn) Date: Thu, 9 Jul 2009 15:33:14 -0400 Subject: vburshteyn In-Reply-To: <20090709143139.GF72073@forum.nginx.org> References: <2c449df47c612bdeaa42d965f6c610af.NginxMailingList@forum.nginx.org> <20090709143139.GF72073@forum.nginx.org> Message-ID: Igor, thanks, bolshoye spasibo. if you dont mind, one more question on that topic - how would u do this? ../ this needs to be redirected to ...// Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3863,3947#msg-3947 From is at rambler-co.ru Thu Jul 9 23:47:30 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Thu, 9 Jul 2009 23:47:30 +0400 Subject: Password-protected dirs working but reporting unexpected errors In-Reply-To: References: <20090709075829.GE66728@rambler-co.ru> <20090709144835.GG72073@rambler-co.ru> Message-ID: <20090709194730.GC80653@rambler-co.ru> On Thu, Jul 09, 2009 at 05:20:54AM -1000, TinyApps.Org wrote: > >>I'd like to send you a donation, but do not see > >>a button on http://sysoev.ru/en/ or http://nginx.net/ . > >>Do the donations to http://wiki.nginx.org/Main really > >>go to you? Just want to make sure. > > >Yes, http://sysoev.ru/en/donation.html > > Thanks - done! Thank you very much! -- Igor Sysoev http://sysoev.ru/en/ From is at rambler-co.ru Thu Jul 9 23:58:20 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Thu, 9 Jul 2009 23:58:20 +0400 Subject: vburshteyn In-Reply-To: References: <2c449df47c612bdeaa42d965f6c610af.NginxMailingList@forum.nginx.org> <20090709143139.GF72073@forum.nginx.org> Message-ID: <20090709195820.GD80653@rambler-co.ru> On Thu, Jul 09, 2009 at 03:33:14PM -0400, vburshteyn wrote: > Igor, thanks, bolshoye spasibo. if you dont mind, one more question on that topic - how would u do this? > > ../ this needs to be redirected to > > ...// > > Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3863,3947#msg-3947 Forum's maillist gateway cut some characters: ---------- [affiliate].[www2].[mysite.com]/[boxoffice][what ever the string in here] this needs to be redirected to [affiliate].[boxoffice].[www2].[mysite.com]/[boxoffice]/[what ever the string in here] ---------- became ---------- ../ this needs to be redirected to ...// ---------- If you want do some generic redirect rule, then server { server_name affiliate.www2.mysite.com; location ~ ^(/[^/]+) { set $name $1; rewrite ^ http://affiliate.$name.www2.mysite.com$request_uri?; } } -- Igor Sysoev http://sysoev.ru/en/ From is at rambler-co.ru Thu Jul 9 23:59:54 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Thu, 9 Jul 2009 23:59:54 +0400 Subject: Auth_basic 500 error In-Reply-To: <4A56322E.6020703@tomvalentine.net> References: <4A56322E.6020703@tomvalentine.net> Message-ID: <20090709195954.GE80653@rambler-co.ru> On Thu, Jul 09, 2009 at 07:08:46PM +0100, nginx at tomvalentine.net wrote: > Hello, > > I can not seem to get auth basic to work, my configuration works fine > with everything else, but if I enable authorisation, I get a 500 > Internal Server Error > > I think the key error is the Readfile() error, but I have no idea what > it means or how to fix it. > > From nginx.conf > > location / { > index index.html index.htm index.php; > auth_basic private; > auth_basic_user_file htpasswd; > } > > If anyone can help, thanks! > By the way, I am using nginx/0.7.61 on Windows Vista, and the error > happens in both Firefox/Opera > 2009/07/09 17:01:45 [error] 1348#504: *3 ReadFile() failed (38: Reached the end of the file), client: 127.0.0.1, server: localhost, request: "GET / HTTP/1.1", host: "localhost:88" This is nginx bug, I will try to reproduce it. How big is your htpasswd ? -- Igor Sysoev http://sysoev.ru/en/ From tom at tomvalentine.net Fri Jul 10 00:28:31 2009 From: tom at tomvalentine.net (Tom Valentine) Date: Thu, 09 Jul 2009 21:28:31 +0100 Subject: Auth_basic 500 error In-Reply-To: References: <4A56322E.6020703@tomvalentine.net> Message-ID: <4A5652EF.1010503@tomvalentine.net> Michael Shadle wrote: > try the full path to the htpasswd file. it's probably trying to read > from the nginx config path. > Hi, I changed it to the full file path, and got the same error. I then changed the configuration to point to a file that did not exist. This resulted in the error: 2009/07/09 21:12:54 [error] 3820#3944: *3 CreateFile() "C:/nginx/conf/.tpasswd" failed (2: The system cannot find the file specified), client: 192.168.1.1, server: localhost, request: "GET /favicon.ico?hj HTTP/1.1", host: "192.168.1.11:88" I pointed nginx.conf back to the correct file and switched to a putty session on my router and used wget for some testing: wget http://:d at 192.168.1.11:88/favicon.ico --> server returned error: HTTP/1.1 401 Unauthorized (Expected) But as soon as I add in a username to the request I got a 500 error: wget http://fdfdf:fsdfsdfsdfsdfsdfd at 192.168.1.11:88/favicon.ico --> server returned error: HTTP/1.1 500 Internal Server Error The only time I got a 403 error was when I was pointing nginx.conf to a non existent file AND if I specified a username in the request: wget http://vps:sdsd at 192.168.1.11:88/favicon.ico --> server returned error: HTTP/1.1 403 Forbidden So it seems I can only get the authorisation to work if I don't have a password file. So is my password file the problem, I generated the passwords from various sources (I tried several) including a working htpasswd file from apache, but I would expect a 403 error, not a 500. From tom at tomvalentine.net Fri Jul 10 00:31:02 2009 From: tom at tomvalentine.net (Tom Valentine) Date: Thu, 09 Jul 2009 21:31:02 +0100 Subject: Auth_basic 500 error In-Reply-To: <20090709195954.GE80653@rambler-co.ru> References: <4A56322E.6020703@tomvalentine.net> <20090709195954.GE80653@rambler-co.ru> Message-ID: <4A565386.8040904@tomvalentine.net> Igor Sysoev wrote: > On Thu, Jul 09, 2009 at 07:08:46PM +0100, nginx at tomvalentine.net wrote: > >> Hello, >> >> I can not seem to get auth basic to work, my configuration works fine >> with everything else, but if I enable authorisation, I get a 500 >> Internal Server Error >> >> I think the key error is the Readfile() error, but I have no idea what >> it means or how to fix it. >> >> From nginx.conf >> >> location / { >> index index.html index.htm index.php; >> auth_basic private; >> auth_basic_user_file htpasswd; >> } >> >> If anyone can help, thanks! >> By the way, I am using nginx/0.7.61 on Windows Vista, and the error >> happens in both Firefox/Opera > >> 2009/07/09 17:01:45 [error] 1348#504: *3 ReadFile() failed (38: Reached the end of the file), client: 127.0.0.1, server: localhost, request: "GET / HTTP/1.1", host: "localhost:88" > > This is nginx bug, I will try to reproduce it. How big is your htpasswd ? > > Currently one line, I had a few more user/password combinations in at one point when I was trying to figure out what was going wrong. Tom From nginx at tomvalentine.net Fri Jul 10 00:44:00 2009 From: nginx at tomvalentine.net (nginx at tomvalentine.net) Date: Thu, 09 Jul 2009 21:44:00 +0100 Subject: Auth_basic 500 error Message-ID: <4A565690.3040703@tomvalentine.net> Igor Sysoev wrote: > On Thu, Jul 09, 2009 at 07:08:46PM +0100, nginx at tomvalentine.net wrote: > >> Hello, >> >> I can not seem to get auth basic to work, my configuration works fine >> with everything else, but if I enable authorisation, I get a 500 >> Internal Server Error >> >> I think the key error is the Readfile() error, but I have no idea what >> it means or how to fix it. >> >> From nginx.conf >> >> location / { >> index index.html index.htm index.php; >> auth_basic private; >> auth_basic_user_file htpasswd; >> } >> >> If anyone can help, thanks! >> By the way, I am using nginx/0.7.61 on Windows Vista, and the error >> happens in both Firefox/Opera > >> 2009/07/09 17:01:45 [error] 1348#504: *3 ReadFile() failed (38: Reached the end of the file), client: 127.0.0.1, server: localhost, request: "GET / HTTP/1.1", host: "localhost:88" > > This is nginx bug, I will try to reproduce it. How big is your htpasswd ? > > Currently one line, I had a few more user/password combinations in at one point when I was trying to figure out what was going wrong. Tom (Sorry, forgot I had different users setup for my email so sent with other id which meant it didn't make it to the mailing list) From cliff at develix.com Fri Jul 10 01:20:26 2009 From: cliff at develix.com (Cliff Wells) Date: Thu, 09 Jul 2009 14:20:26 -0700 Subject: Nginx proxying WebDAV = no love In-Reply-To: <20090709070009.GA66728@rambler-co.ru> References: <1247087646.3689.9.camel@portable-evil> <20090709070009.GA66728@rambler-co.ru> Message-ID: <1247174426.3689.49.camel@portable-evil> On Thu, 2009-07-09 at 11:00 +0400, Igor Sysoev wrote: > Coud you create debug log of the requests ? Sent via private email. Regards, Cliff -- http://www.google.com/search?q=vonage+sucks From nbubingo at gmail.com Fri Jul 10 06:34:48 2009 From: nbubingo at gmail.com (Weibin Yao) Date: Fri, 10 Jul 2009 10:34:48 +0800 Subject: Upstream (fair) with dead servers In-Reply-To: <1c14c1730907090747w4294ecffm3d1fc3d3c9010b0f@mail.gmail.com> References: <1c14c1730907090747w4294ecffm3d1fc3d3c9010b0f@mail.gmail.com> Message-ID: <4A56A8C8.1000502@gmail.com> Kamil Gorlo at 2009-7-9 22:47 wrote: > Hi, > > I have (possibly quite rare) situation where I have one upstream with > two servers and I know that in any particular moment only one of those > servers is running: > > upstream blah { > server server1; // when working, server2 is dead > server server2; // when working, server1 is dead > } > > So, what I need is that Nginx will choose working backend (in this > upstream), remembers it and choose it for every request till some > failure on this server occurs. Then it tries other server in upstream, > till it finds some working - and situation repeats. Is this possible? > > Maybe upstream_fair module with option no_rr will work for me (I am not sure)? > > Cheers, > I think you should try the keepalived. -- Weibin Yao From nbubingo at gmail.com Fri Jul 10 06:59:12 2009 From: nbubingo at gmail.com (Weibin Yao) Date: Fri, 10 Jul 2009 10:59:12 +0800 Subject: Upstream (fair) with dead servers In-Reply-To: <20090709151454.GI72073@rambler-co.ru> References: <1c14c1730907090747w4294ecffm3d1fc3d3c9010b0f@mail.gmail.com> <20090709145701.GH72073@rambler-co.ru> <1c14c1730907090809r66e6c2abm94a5f2a74bc7079a@mail.gmail.com> <20090709151454.GI72073@rambler-co.ru> Message-ID: <4A56AE80.1050906@gmail.com> Igor Sysoev at 2009-7-9 23:14 wrote: > I'm going to rewrite upstream modules in 0.8.x (add fair, etc.) and > probably add this functionality. But why do you want to send request > to one only server leaving the other as spare one ? > I'm glad to hear the good news, thanks for you hard work. Will you add upstream keepalive in 0.8.X? -- Weibin Yao From is at rambler-co.ru Fri Jul 10 11:07:35 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Fri, 10 Jul 2009 11:07:35 +0400 Subject: Nginx proxying WebDAV = no love In-Reply-To: <1247174426.3689.49.camel@portable-evil> References: <1247087646.3689.9.camel@portable-evil> <20090709070009.GA66728@rambler-co.ru> <1247174426.3689.49.camel@portable-evil> Message-ID: <20090710070735.GD34030@rambler-co.ru> On Thu, Jul 09, 2009 at 02:20:26PM -0700, Cliff Wells wrote: > On Thu, 2009-07-09 at 11:00 +0400, Igor Sysoev wrote: > > > Coud you create debug log of the requests ? > > Sent via private email. As it was suggested before, the backend sends a chunked response to nginx's HTTP/1.0 request: .. 207 Multi-Status Server: DAV/0.9.3 Python/2.6.2 Connection: close Date: Thu, 09 Jul 2009 21:17:01 GMT Content-type: text/xml; encoding="utf-8" Connection: close Transfer-Encoding: chunked Date: Thu, 09 Jul 2009 21:17:01 GMT -- Igor Sysoev http://sysoev.ru/en/ From dave at cheney.net Fri Jul 10 11:43:30 2009 From: dave at cheney.net (Dave Cheney) Date: Fri, 10 Jul 2009 03:43:30 -0400 Subject: Nginx proxying WebDAV =?UTF-8?B?PQ==?= no love References: <1247087646.3689.9.camel@portable-evil> <20090709070009.GA66728@rambler-co.ru> <1247174426.3689.49.camel@portable-evil> <20090710070735.GD34030@rambler-co.ru> Message-ID: What I found was the OS X as the client would send a chunked *request* which nginx does not support, however sending a chunked response to a HTTP/1.0 request is bad form as well. Cheers Dave Igor Sysoev writes: > On Thu, Jul 09, 2009 at 02:20:26PM -0700, Cliff Wells wrote: > >> On Thu, 2009-07-09 at 11:00 +0400, Igor Sysoev wrote: >> >> > Coud you create debug log of the requests ? >> >> Sent via private email. > > As it was suggested before, the backend sends a chunked response > to nginx's HTTP/1.0 request: > > .. 207 Multi-Status > Server: DAV/0.9.3 Python/2.6.2 > Connection: close > Date: Thu, 09 Jul 2009 21:17:01 GMT > Content-type: text/xml; encoding="utf-8" > Connection: close > Transfer-Encoding: chunked > Date: Thu, 09 Jul 2009 21:17:01 GMT > > > -- > Igor Sysoev > http://sysoev.ru/en/ > From is at rambler-co.ru Fri Jul 10 11:55:44 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Fri, 10 Jul 2009 11:55:44 +0400 Subject: Nginx proxying WebDAV = no love In-Reply-To: References: <1247087646.3689.9.camel@portable-evil> <20090709070009.GA66728@rambler-co.ru> <1247174426.3689.49.camel@portable-evil> <20090710070735.GD34030@rambler-co.ru> Message-ID: <20090710075544.GE34030@rambler-co.ru> On Fri, Jul 10, 2009 at 03:43:30AM -0400, Dave Cheney wrote: > What I found was the OS X as the client would send a chunked *request* > which nginx does not support, however sending a chunked response to a > HTTP/1.0 request is bad form as well. You are right, I have incorrectly interpreted your sentence: --- Get out ngrep and make sure that pywebdav is honoring the Connection: close request proxied via nginx. I have found that nginx waits for the --- > Cheers > > Dave > > > Igor Sysoev writes: > > >On Thu, Jul 09, 2009 at 02:20:26PM -0700, Cliff Wells wrote: > > > >>On Thu, 2009-07-09 at 11:00 +0400, Igor Sysoev wrote: > >> > >>> Coud you create debug log of the requests ? > >> > >>Sent via private email. > > > >As it was suggested before, the backend sends a chunked response > >to nginx's HTTP/1.0 request: > > > >.. 207 Multi-Status > >Server: DAV/0.9.3 Python/2.6.2 > >Connection: close > >Date: Thu, 09 Jul 2009 21:17:01 GMT > >Content-type: text/xml; encoding="utf-8" > >Connection: close > >Transfer-Encoding: chunked > >Date: Thu, 09 Jul 2009 21:17:01 GMT > > > > > >-- > >Igor Sysoev > >http://sysoev.ru/en/ > > -- Igor Sysoev http://sysoev.ru/en/ From is at rambler-co.ru Fri Jul 10 14:20:24 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Fri, 10 Jul 2009 14:20:24 +0400 Subject: Upstream (fair) with dead servers In-Reply-To: <4A56AE80.1050906@gmail.com> References: <1c14c1730907090747w4294ecffm3d1fc3d3c9010b0f@mail.gmail.com> <20090709145701.GH72073@rambler-co.ru> <1c14c1730907090809r66e6c2abm94a5f2a74bc7079a@mail.gmail.com> <20090709151454.GI72073@rambler-co.ru> <4A56AE80.1050906@gmail.com> Message-ID: <20090710102024.GC38566@rambler-co.ru> On Fri, Jul 10, 2009 at 10:59:12AM +0800, Weibin Yao wrote: > Igor Sysoev at 2009-7-9 23:14 wrote: > >I'm going to rewrite upstream modules in 0.8.x (add fair, etc.) and > >probably add this functionality. But why do you want to send request > >to one only server leaving the other as spare one ? > > > I'm glad to hear the good news, thanks for you hard work. Will you add > upstream keepalive in 0.8.X? At least I plan this. -- Igor Sysoev http://sysoev.ru/en/ From eduveks at gmail.com Fri Jul 10 18:20:37 2009 From: eduveks at gmail.com (Eduardo Velasques) Date: Fri, 10 Jul 2009 15:20:37 +0100 Subject: Nginx+WebDav vs SVN Message-ID: Hi, Nginx + WebDav works with SVN? Without no proxy to Apache + WebDav? I did an search and I see it is not possible... but is true? Thank you. -------------- next part -------------- An HTML attachment was scrubbed... URL: From mike503 at gmail.com Fri Jul 10 19:44:10 2009 From: mike503 at gmail.com (Michael Shadle) Date: Fri, 10 Jul 2009 08:44:10 -0700 Subject: Nginx+WebDav vs SVN In-Reply-To: References: Message-ID: It does not support it, it is on my wishlist and some others though :) 2009/7/10 Eduardo Velasques : > Hi, > > Nginx + WebDav works with SVN? Without no proxy to Apache + WebDav? > > I did an search and I see it is not possible... but is true? > > Thank you. > > From nginx-forum at nginx.us Fri Jul 10 19:56:33 2009 From: nginx-forum at nginx.us (vburshteyn) Date: Fri, 10 Jul 2009 11:56:33 -0400 Subject: vburshteyn In-Reply-To: <20090709195820.GD80653@forum.nginx.org> References: <2c449df47c612bdeaa42d965f6c610af.NginxMailingList@forum.nginx.org> <20090709143139.GF72073@forum.nginx.org> <20090709195820.GD80653@forum.nginx.org> Message-ID: <358f56bb81dc3ec3e3add921137e650e.NginxMailingList@forum.nginx.org> Igor, in your opinion, if redirects are just say 10% of the traffic, and rest does not need to be redirected, would you just use IF statment? say if (host ~ group) rewrite xxxxxxxxxxxxxxxx or u would still go with virtual server set up that you suggested earlier? Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3863,3968#msg-3968 From is at rambler-co.ru Fri Jul 10 20:01:45 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Fri, 10 Jul 2009 20:01:45 +0400 Subject: vburshteyn In-Reply-To: <358f56bb81dc3ec3e3add921137e650e.NginxMailingList@forum.nginx.org> References: <2c449df47c612bdeaa42d965f6c610af.NginxMailingList@forum.nginx.org> <20090709143139.GF72073@forum.nginx.org> <20090709195820.GD80653@forum.nginx.org> <358f56bb81dc3ec3e3add921137e650e.NginxMailingList@forum.nginx.org> Message-ID: <20090710160145.GD44171@rambler-co.ru> On Fri, Jul 10, 2009 at 11:56:33AM -0400, vburshteyn wrote: > Igor, > > in your opinion, if redirects are just say 10% of the traffic, and rest does not need to be redirected, would you just use IF statment? > > say > > if (host ~ group) > rewrite xxxxxxxxxxxxxxxx > > or u would still go with virtual server set up that you suggested earlier? If you want to write "if (host ~ group)", this means that you should forget about RewriteCond and use "server { server_name group; ... } -- Igor Sysoev http://sysoev.ru/en/ From cliff at develix.com Fri Jul 10 22:53:23 2009 From: cliff at develix.com (Cliff Wells) Date: Fri, 10 Jul 2009 11:53:23 -0700 Subject: Nginx proxying WebDAV = no love In-Reply-To: <20090710070735.GD34030@rambler-co.ru> References: <1247087646.3689.9.camel@portable-evil> <20090709070009.GA66728@rambler-co.ru> <1247174426.3689.49.camel@portable-evil> <20090710070735.GD34030@rambler-co.ru> Message-ID: <1247252003.3689.56.camel@portable-evil> On Fri, 2009-07-10 at 11:07 +0400, Igor Sysoev wrote: > On Thu, Jul 09, 2009 at 02:20:26PM -0700, Cliff Wells wrote: > > > On Thu, 2009-07-09 at 11:00 +0400, Igor Sysoev wrote: > > > > > Coud you create debug log of the requests ? > > > > Sent via private email. > > As it was suggested before, the backend sends a chunked response > to nginx's HTTP/1.0 request: > > .. 207 Multi-Status > Server: DAV/0.9.3 Python/2.6.2 > Connection: close > Date: Thu, 09 Jul 2009 21:17:01 GMT > Content-type: text/xml; encoding="utf-8" > Connection: close > Transfer-Encoding: chunked > Date: Thu, 09 Jul 2009 21:17:01 GMT Thanks Igor. I've fixed it and submitted a patch to the pywebdav developers. Regards, Cliff -- http://www.google.com/search?q=vonage+sucks From apseudoutopia at gmail.com Sat Jul 11 00:15:47 2009 From: apseudoutopia at gmail.com (APseudoUtopia) Date: Fri, 10 Jul 2009 16:15:47 -0400 Subject: Nginx+WebDav vs SVN In-Reply-To: References: Message-ID: <27ade5280907101315x650d086cibf3a07d045a052f1@mail.gmail.com> On Fri, Jul 10, 2009 at 11:44 AM, Michael Shadle wrote: > It does not support it, it is on my wishlist and some others though :) > > 2009/7/10 Eduardo Velasques : >> Hi, >> >> Nginx + WebDav works with SVN? Without no proxy to Apache + WebDav? >> >> I did an search and I see it is not possible... but is true? >> >> Thank you. >> >> > > Running svnserve is probably the next best thing. It's a very lightweight daemon, and it eliminates the need for having apache installed/running (and it seems to be faster too!) From mike503 at gmail.com Sat Jul 11 01:30:36 2009 From: mike503 at gmail.com (Michael Shadle) Date: Fri, 10 Jul 2009 14:30:36 -0700 Subject: Nginx+WebDav vs SVN In-Reply-To: <27ade5280907101315x650d086cibf3a07d045a052f1@mail.gmail.com> References: <27ade5280907101315x650d086cibf3a07d045a052f1@mail.gmail.com> Message-ID: ssh+svn is neat too, you can use ssh keys for auth, and it's encrypted too. only needs a single user account on the svn server too. On Fri, Jul 10, 2009 at 1:15 PM, APseudoUtopia wrote: > On Fri, Jul 10, 2009 at 11:44 AM, Michael Shadle wrote: >> It does not support it, it is on my wishlist and some others though :) >> >> 2009/7/10 Eduardo Velasques : >>> Hi, >>> >>> Nginx + WebDav works with SVN? Without no proxy to Apache + WebDav? >>> >>> I did an search and I see it is not possible... but is true? >>> >>> Thank you. >>> >>> >> >> > > Running svnserve is probably the next best thing. It's a very > lightweight daemon, and it eliminates the need for having apache > installed/running (and it seems to be faster too!) > > From kgorlo at gmail.com Sat Jul 11 02:00:32 2009 From: kgorlo at gmail.com (Kamil Gorlo) Date: Sat, 11 Jul 2009 00:00:32 +0200 Subject: Upstream (fair) with dead servers In-Reply-To: <4A56A8C8.1000502@gmail.com> References: <1c14c1730907090747w4294ecffm3d1fc3d3c9010b0f@mail.gmail.com> <4A56A8C8.1000502@gmail.com> Message-ID: <1c14c1730907101500j241c5490o27e8e168ebe4bb65@mail.gmail.com> On Fri, Jul 10, 2009 at 4:34 AM, Weibin Yao wrote: > Kamil Gorlo at 2009-7-9 22:47 wrote: >> >> Hi, >> >> I have (possibly quite rare) situation where I have one upstream with >> two servers and I know that in any particular moment only one of those >> servers is running: >> >> upstream blah { >> ? ?server server1; // when working, server2 is dead >> ? ?server server2; // when working, server1 is dead >> } >> >> So, what I need is that Nginx will choose working backend (in this >> upstream), remembers it and choose it for every request till some >> failure on this server occurs. Then it tries other server in upstream, >> till it finds some working - and situation repeats. Is this possible? >> >> Maybe upstream_fair module with option no_rr will work for me (I am not >> sure)? >> >> Cheers, >> > > I think you should try the keepalived. What do you mean? Cheers, Kamil From merlincorey at dc949.org Sat Jul 11 02:22:59 2009 From: merlincorey at dc949.org (merlin corey) Date: Fri, 10 Jul 2009 15:22:59 -0700 Subject: Nginx+WebDav vs SVN In-Reply-To: References: <27ade5280907101315x650d086cibf3a07d045a052f1@mail.gmail.com> Message-ID: Even better, in my opinion, and long the same vein, is git, which uses a single user account with SSH key access to multiple users/repositories (check out gitosis for a nice management setup using git itself). On Fri, Jul 10, 2009 at 2:30 PM, Michael Shadle wrote: > ssh+svn is neat too, you can use ssh keys for auth, and it's encrypted > too. only needs a single user account on the svn server too. > > On Fri, Jul 10, 2009 at 1:15 PM, APseudoUtopia wrote: >> On Fri, Jul 10, 2009 at 11:44 AM, Michael Shadle wrote: >>> It does not support it, it is on my wishlist and some others though :) >>> >>> 2009/7/10 Eduardo Velasques : >>>> Hi, >>>> >>>> Nginx + WebDav works with SVN? Without no proxy to Apache + WebDav? >>>> >>>> I did an search and I see it is not possible... but is true? >>>> >>>> Thank you. >>>> >>>> >>> >>> >> >> Running svnserve is probably the next best thing. It's a very >> lightweight daemon, and it eliminates the need for having apache >> installed/running (and it seems to be faster too!) >> >> > > From mike503 at gmail.com Sat Jul 11 03:08:29 2009 From: mike503 at gmail.com (Michael Shadle) Date: Fri, 10 Jul 2009 16:08:29 -0700 Subject: Nginx+WebDav vs SVN In-Reply-To: References: <27ade5280907101315x650d086cibf3a07d045a052f1@mail.gmail.com> Message-ID: it's basically the same method using svn+ssh the way i'm using it (i've used gitosis too, it essentially just manages your authorized_keys file with the right command line stuff) so this gives the same benefits as gitosis. i wonder if anyone has svnosis out there :) On Fri, Jul 10, 2009 at 3:22 PM, merlin corey wrote: > Even better, in my opinion, and long the same vein, is git, which uses > a single user account with SSH key access to multiple > users/repositories (check out gitosis for a nice management setup > using git itself). > > On Fri, Jul 10, 2009 at 2:30 PM, Michael Shadle wrote: >> ssh+svn is neat too, you can use ssh keys for auth, and it's encrypted >> too. only needs a single user account on the svn server too. From jim.ohlstein at gmail.com Sat Jul 11 03:39:46 2009 From: jim.ohlstein at gmail.com (Jim Ohlstein) Date: Fri, 10 Jul 2009 19:39:46 -0400 Subject: Cache questions Message-ID: <4A57D142.2070501@gmail.com> I'm using the fastcgi cache for static files (images, javascript,css) and just found multiple lines in the error log like this one: 2009/07/10 10:22:54 [crit] 22476#0: ngx_slab_alloc() failed: no memory in cache keys zone "one" So I increased the memory available for the zone and reloaded nginx. It took over five hours to go through the cache but these are the relevant entries: 2009/07/10 12:11:03 [notice] 21038#0: start cache manager process 32730 2009/07/10 12:11:04 [notice] 21038#0: cache manager process 22480 exited with code 0 and finally 2009/07/10 17:43:27 [notice] 32730#0: http file cache: /usr/local/nginx/cache 11638.289M, bsize: 4096 My questions are: Is that simply the total (11638.289MB or 11.4GB) of all of the file sizes, or is that the actual disk space consumed taking into account total number of blocks used multiplied by the block size? The number could be quite different given what I estimate are nearly one million mostly small files in the cache at this point. When I next upgrade nginx (I'm running 0.8.4), and I attempt a "graceful upgrade" will it have to go through this entire process again? From lists at ruby-forum.com Sat Jul 11 10:03:58 2009 From: lists at ruby-forum.com (Qaiser Wali) Date: Sat, 11 Jul 2009 08:03:58 +0200 Subject: Serving Dynamic js files Message-ID: <3cdc0eccaf279b05a0ec7467c74201aa@ruby-forum.com> I have a controller named javascripts and in my view i have few .js.erb files. on my local machine, i can include the .js as a regular script file and it works fine. on my production machine i am not sure how to configure nginx that it can recognize both the folders public/javascripts and apps/views/javascripts and serve javascripts from both these places. someone help please. -- Posted via http://www.ruby-forum.com/. From merlincorey at dc949.org Sat Jul 11 10:58:55 2009 From: merlincorey at dc949.org (merlin corey) Date: Fri, 10 Jul 2009 23:58:55 -0700 Subject: Nginx+WebDav vs SVN In-Reply-To: References: <27ade5280907101315x650d086cibf3a07d045a052f1@mail.gmail.com> Message-ID: The nice thing about gitosis over svnadmin is that gitosis itself is in a repository and does it's work with post-commit-hooks so you can have a couple people able to manage it. However, I agree that essentially it is the same situation for developers as svn+ssh (aside from the extra branching capabilities of git and decentralized nature) which is why I said it was in the same vein :P. On Fri, Jul 10, 2009 at 4:08 PM, Michael Shadle wrote: > it's basically the same method using svn+ssh the way i'm using it > (i've used gitosis too, it essentially just manages your > authorized_keys file with the right command line stuff) > > so this gives the same benefits as gitosis. i wonder if anyone has > svnosis out there :) > > On Fri, Jul 10, 2009 at 3:22 PM, merlin corey wrote: >> Even better, in my opinion, and long the same vein, is git, which uses >> a single user account with SSH key access to multiple >> users/repositories (check out gitosis for a nice management setup >> using git itself). >> >> On Fri, Jul 10, 2009 at 2:30 PM, Michael Shadle wrote: >>> ssh+svn is neat too, you can use ssh keys for auth, and it's encrypted >>> too. only needs a single user account on the svn server too. > > From nginx-list at deepstorm.com Sat Jul 11 11:28:05 2009 From: nginx-list at deepstorm.com (Shawn Grunberger) Date: Sat, 11 Jul 2009 00:28:05 -0700 Subject: Setting headers on subrequests Message-ID: <596495cd0907110028w49d87935n8d23775c8c956b@mail.gmail.com> I'm investigating a bug in Evan Miller's excellent mod_zip add-on. The mod_zip code, running as a filter, issues one or more subrequests. To support byte ranges, Evan sends a different Range header with each subrequest (thus relying on nginx's built-in Range filter to do the actual truncation). To accomplish this, the code modifies the r->headers_in.range structure of the main request before each call to ngx_http_subrequest (the code calls this "a dirty hack"). In psuedocode it looks like this: while (rc == NGX_OK && more_subrequests_to_make) { modify r->headers_in.range rc = ngx_http_subrequest(r, uri, args, &sr, NULL, 0) } return rc Assume that each call to ngx_http_subrequest returns NGX_OK (in other words, the subrequests require no network access). In that case, this code works in nginx 0.7.24 and below. A subrequest's headers are committed and sent immediately after the call to ngx_http_subrequest. In 0.7.25 and above, however, this code fails to set a unique header per subrequest. Instead, all subrequests inherit the Range header set in the final loop iteration. In other words, no subrequest is committed and sent until after the loop completes. Is there a "proper" way to set different header values per subrequest? From antonio04 at gmail.com Sat Jul 11 12:05:59 2009 From: antonio04 at gmail.com (Antonio L.) Date: Sat, 11 Jul 2009 01:05:59 -0700 Subject: Dramatic slowdown on increased connections (fbsd/nginx/php-fpm) Message-ID: Hi folks -- I apologize in advance, as this isn't necessarily a clear-cut Nginx issue per se, but I'd appreciate any pointers that anyone might have. I'm transitioning a web site with app servers currently running FreeBSD/Apache/mod_php into a new setup where they will run nginx+php-fpm. I have been testing with a machine that is running FreeBSD 7.1-Release, nginx/0.7.60, PHP 5.2.8/fpm-0.5.10. It's a dual quad-core (Xeon 5405) with 12GB. I offloaded some live traffic from our app servers to this test server to see how it performed, gradually increasing the load. We got to around 250 reqs/s (measured using the nginx stub status module -- and it's purely dynamic traffic -- static stuff is served separately) with no strain on the server (avg load on procs was < 1, no disk blocking in vmstat, etc.). As soon as we passed the 250 req/s mark, we witnessed an instant slowdown. On the nginx status page, "Writing: " jumps from approx. 15-20 to over 2000 (continuing to pile up rapidly). However, server load stays constant and I didn't notice significant I/O changes (in vmstat/iostat) -- but maybe I missed something. Likewise, kern.openfiles rose to maybe ~ 2000 (well below the limit) and kern.ipc.numopensockets was similar. This occurs if we keep the load constant at around 200 reqs/s, just not as fast -- after a couple of hours, though, the same thing happens -- and traffic just piles up. The only way we can resolve it is to restart nginx / php-fpm and decrease the traffic load sent to the server. This is what my /boot/loader.conf looks like: kern.maxdsiz="11264M" kern.dfldsiz="768M" kern.maxssiz="256MB" vm.kmem_size="1400M" kern.maxbcache="64M" kern.ipc.maxpipekva="4M" kern.maxfiles="65536" kern.ipc.nmbclusters="65536" And /etc/sysctl.conf contains: kern.ipc.somaxconn=8192 kern.ipc.nmbclusters=32768 In php-fpm.conf I've got rlimit_files set to 30000 and max_children to 200. Does anyone have any advice on how to further investigate this issue? I would expect the load on the server to gradually increase, not pile up like this, so I assume some system or software configuration value must be holding this back. Thanks! Antonio -------------- next part -------------- An HTML attachment was scrubbed... URL: From lists at ruby-forum.com Sat Jul 11 19:13:20 2009 From: lists at ruby-forum.com (Andreas Vieten) Date: Sat, 11 Jul 2009 17:13:20 +0200 Subject: Can't start nginx Message-ID: <808d0ac45d2eeeabbc7dbf3881810c7b@ruby-forum.com> Hello, I've tried a serveral times to start nginx but it always ends with no happy end. Download and compiling is no problem for me. So here goes my way that i do it: 1) cd /tmp wget wget http://sysoev.ru/nginx/nginx-0.7.61.tar.gz tar xvfz nginx-0.7.61.tar.gz cd nginx-0.7.61 ./configure \ --prefix=/usr \ --conf-path=/etc/nginx/nginx.conf \ --http-log-path=/var/log/nginx/access_log \ --error-log-path=/var/log/nginx/error_log \ --pid-path=/var/run/nginx.pid \ --http-client-body-temp-path=/var/tmp/nginx/client \ --http-proxy-temp-path=/var/tmp/nginx/proxy \ --http-fastcgi-temp-path=/var/tmp/nginx/fastcgi \ --with-md5-asm --with-md5=/usr/include \ --with-http_realip_module \ --with-http_ssl_module \ --with-http_stub_status_module make make install 2) vi /etc/init.d/nginx #! /bin/sh ### BEGIN INIT INFO # Provides: nginx # Required-Start: $all # Required-Stop: $all # Default-Start: 2 3 4 5 # Default-Stop: 0 1 6 # Short-Description: starts the nginx web server # Description: starts nginx using start-stop-daemon ### END INIT INFO PATH=/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin DAEMON=/usr/local/sbin/nginx NAME=nginx DESC=nginx test -x $DAEMON || exit 0 # Include nginx defaults if available if [ -f /etc/default/nginx ] ; then . /etc/default/nginx fi set -e case "$1" in start) echo -n "Starting $DESC: " start-stop-daemon --start --quiet --pidfile /usr/local/nginx/logs/nginx.pid --exec $DAEMON -- $DAEMON_OPTS echo "$NAME." ;; stop) echo -n "Stopping $DESC: " start-stop-daemon --stop --quiet --pidfile /usr/local/nginx/logs/nginx.pid --exec $DAEMON echo "$NAME." ;; restart|force-reload) echo -n "Restarting $DESC: " start-stop-daemon --stop --quiet --pidfile /usr/local/nginx/logs/nginx.pid --exec $DAEMON sleep 1 start-stop-daemon --start --quiet --pidfile /usr/local/nginx/logs/nginx.pid --exec $DAEMON -- $DAEMON_OPTS echo "$NAME." ;; reload) echo -n "Reloading $DESC configuration: " start-stop-daemon --stop --signal HUP --quiet --pidfile /usr/local/nginx/logs/nginx.pid --exec $DAEMON echo "$NAME." ;; *) N=/etc/init.d/$NAME echo "Usage: $N {start|stop|restart|force-reload}" >&2 exit 1 ;; esac exit 0 3) chmod 755 /etc/init.d/nginx 4) /etc/init.d/nginx start But it won't start because of this error here: (root)-(/tmp/nginx-0.7.61) $ /etc/init.d/nginx start /etc/init.d/nginx: line 1: N: command not found -- Posted via http://www.ruby-forum.com/. From mikemc-nginx at terabytemedia.com Sat Jul 11 20:28:01 2009 From: mikemc-nginx at terabytemedia.com (Michael McCallister) Date: Sat, 11 Jul 2009 09:28:01 -0700 Subject: behaviour changes upgrading from nginx .6.x to .7.x Message-ID: <4A58BD91.3090909@terabytemedia.com> Greetings, First, thanks for nginx - very cool web server. Next, upon upgrading from 0.6.36 to 0.7.61 I noticed some changes in behaviour which prompted me to downgrade back to 0.6.36 until I could further understand the changes between .6.x and .7.x. More specifically, here is where things worked differently: Given this rewrite rule: rewrite ^/[Aa]rticles/?(.*) /search.php?q=$1; This URL: /Articles/highway+diesel+vehicles Would generate the following $_GET in php (5.2.8 using fpm) and nginx (0.6.36): array ( 'q' => 'highway diesel vehicles', ) Whereas the same exact URL would generate the following $_GET in php (5.2.8 using fpm - no changes to php) and nginx (0.7.61 - nginx is the only thing that changed): array ( 'q' => 'highway+diesel+vehicles', ) I view the behaviour under nginx 0.6.36 as correct while the behaviour under 0.7.61 does not result in the expected value. Obviously it is trivial to workaround this issue, but this leads up to my main concern - what else behaves differently? I tried to find documentation outlining upgrade considerations while migrating from .6.x to .7.x but could not find any - my apologies if I missed them and/or am asking a question already addressed previously on the list. So I guess I really have two questions considering the above: 1) Is the above behaviour in .7.x expected? 2) Is there a list of things to watch out for when changing from .6.x to .7.x? Any help is greatly appreciated and thanks again for nginx. Michael From cliff at develix.com Sat Jul 11 22:35:15 2009 From: cliff at develix.com (Cliff Wells) Date: Sat, 11 Jul 2009 11:35:15 -0700 Subject: Serving Dynamic js files In-Reply-To: <3cdc0eccaf279b05a0ec7467c74201aa@ruby-forum.com> References: <3cdc0eccaf279b05a0ec7467c74201aa@ruby-forum.com> Message-ID: <1247337315.3689.67.camel@portable-evil> On Sat, 2009-07-11 at 08:03 +0200, Qaiser Wali wrote: > I have a controller named javascripts and in my view i have few .js.erb > files. > > on my local machine, i can include the .js as a regular script file and > it works fine. on my production machine i am not sure how to configure > nginx that it can recognize both the folders public/javascripts and > apps/views/javascripts and serve javascripts from both these places. Use two locations. location /public/javascripts { root /someplace/public/javascripts; } location /views/javascripts { proxy_pass http://127.0.0.1; # proxy to rails } Then just refer to the two urls in your templates. Cliff -- http://www.google.com/search?q=vonage+sucks From nginx-forum at nginx.us Mon Jul 13 00:13:17 2009 From: nginx-forum at nginx.us (compute) Date: Sun, 12 Jul 2009 16:13:17 -0400 Subject: Howto update to the latest version? Message-ID: Hello, i've installed nginx on debian via But it is not the latest version it is only a old one So how can i update for example to the latest version? Best regards. Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3994,3994#msg-3994 From almir at almirkaric.com Mon Jul 13 00:55:51 2009 From: almir at almirkaric.com (Almir Karic) Date: Sun, 12 Jul 2009 13:55:51 -0700 Subject: Howto update to the latest version? In-Reply-To: References: Message-ID: <2b6a8c330907121355j67dd4a97lf318087d08dd3f85@mail.gmail.com> why do you want to update? if there is no feature you need from new versions i'd advise to stay with what debian provides. if you insist on upgrading, you will have to download the source, compile it and install :-), something ala http://www.softwareprojects.com/resources/programming/t-recompileupgrade-nginx-binary-with-no-down-time-1520.html only that for /etc/init.d/nginx to work as expected you will have to change the paths which are passed to ./configure. On Sun, Jul 12, 2009 at 1:13 PM, compute wrote: > Hello, > > i've installed nginx on debian via > > But it is not the latest version it is only a old one > > > So how can i update for example to the latest version? > > Best regards. > > Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3994,3994#msg-3994 > > > -- python/django hacker & sys admin almirkaric.com & twitter.com/redduck666 From merlincorey at dc949.org Mon Jul 13 05:51:42 2009 From: merlincorey at dc949.org (merlin corey) Date: Sun, 12 Jul 2009 18:51:42 -0700 Subject: Howto update to the latest version? In-Reply-To: <2b6a8c330907121355j67dd4a97lf318087d08dd3f85@mail.gmail.com> References: <2b6a8c330907121355j67dd4a97lf318087d08dd3f85@mail.gmail.com> Message-ID: You can get the current configure string with nginx -V at the commandline. This makes replacing nginx binaries a breeze. For your debian related questions, you should really be looking for debian related resources. It sounds like perhaps you are on etch with nginx 0.5.x (though you did not say) and if that is the case, I agree with the above poster - *if it ain't broke DON'T fix it* . The "debian way" is for stable branches to keep the packages they were given - this keeps them stable! If you need a newer version of software on debian (the debian way), you either need to get a "backport", make your own package, or update to a newer release off the operating system. -- Merlin On Sun, Jul 12, 2009 at 1:55 PM, Almir Karic wrote: > why do you want to update? if there is no feature you need from new > versions i'd advise to stay with what debian provides. > > if you insist on upgrading, you will have to download the source, > compile it and install :-), something ala > http://www.softwareprojects.com/resources/programming/t-recompileupgrade-nginx-binary-with-no-down-time-1520.html > only that for /etc/init.d/nginx to work as expected you will have to > change the paths which are passed to ./configure. > > > On Sun, Jul 12, 2009 at 1:13 PM, compute wrote: >> Hello, >> >> i've installed nginx on debian via >> >> But it is not the latest version it is only a old one >> >> >> So how can i update for example to the latest version? >> >> Best regards. >> >> Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3994,3994#msg-3994 >> >> >> > > > > -- > python/django hacker & sys admin > almirkaric.com & twitter.com/redduck666 > > From merlincorey at dc949.org Mon Jul 13 11:17:02 2009 From: merlincorey at dc949.org (merlin corey) Date: Mon, 13 Jul 2009 00:17:02 -0700 Subject: SSL server_name support, single server block plain/SSL, et al in 0.8.x ? Message-ID: Hello, I had a long chat with a user in the IRC channel about SSL in nginx. Initially, it started out talking about hosting multiple domains with SSL on the same address. They pointed me to http://tools.ietf.org/html/rfc4366#section-3.1 which seems fairly straightforward and I am curious if there is interest in this or if Igor plans to implement it at some point in future anyway. Also, through talking with him, he showed me his idea for how the SSL support should work. It went more or less like this: ssl_certificate /etc/ssl/custom/supercatchall.crt; ssl_certificate_key /etc/ssl/custom/supercatchall.key; server { listen 80; listen 443; if(port = 443) { ssl on; } ... normal server config ... } ... repeated for several servers apparently on same IP ... I tried to talk him out of using if and using a separate server block for 443 and 80 ports for each server, and just including common configuration. Apparently, this does not currently work, but it seems it should be more or less usable, I think he was having problems with using default ssl in the listen line and NginX would not bind multiple times. At any rate, he also complained about the "roundabout redundancy" of the config. That got me to thinking, would something like the following be desirable, and how difficult to implement would it be? server { listen 80; listen 443; ssl on 443; ... normal server config ... } My gut instinct tells me this would be a lot more work than simply allowing multiple SSL hosts per IP address, but it does seem to have a nice ring to it, this late in the day. So to summarize, I am curious if 0.8.x is planned to support multiple SSL hosts per IP, if that feature is even desired by anyone (or maybe other ways to do it, in case I see this problem again!), and finally anyone and everyone's thoughts on the above syntax to unify config for SSL and non SSL (you could still use separate blocks if the configuration is not exactly the same for plain and SSL, of course). Thanks, Merlin From almir at almirkaric.com Mon Jul 13 11:53:41 2009 From: almir at almirkaric.com (Almir Karic) Date: Mon, 13 Jul 2009 00:53:41 -0700 Subject: SSL server_name support, single server block plain/SSL, et al in 0.8.x ? In-Reply-To: References: Message-ID: <2b6a8c330907130053h5a87e8dcl851442ad505420f@mail.gmail.com> You can't do multiple SSL hosts on an ip :-), as you can see on the top of the document you linked to that is for TLS. Nginx already knows SNI. IIRC the issue is if client supports SNI. As for the if(), you don't need it: http://serverfault.com/questions/10854/nginx-https-serving-with-same-config-as-http :-) python/django hacker & sys admin http://almirkaric.com & http://twitter.com/redduck666 On Mon, Jul 13, 2009 at 12:17 AM, merlin corey wrote: > Hello, > > I had a long chat with a user in the IRC channel about SSL in nginx. > Initially, it started out talking about hosting multiple domains with > SSL on the same address. ?They pointed me to > http://tools.ietf.org/html/rfc4366#section-3.1 which seems fairly > straightforward and I am curious if there is interest in this or if > Igor plans to implement it at some point in future anyway. > > Also, through talking with him, he showed me his idea for how the SSL > support should work. ?It went more or less like this: > > ssl_certificate ? ? ?/etc/ssl/custom/supercatchall.crt; > ssl_certificate_key ?/etc/ssl/custom/supercatchall.key; > server { > ?listen 80; > ?listen 443; > ?if(port = 443) { > ? ?ssl on; > ?} > ?... normal server config ... > } > ... repeated for several servers apparently on same IP ... > > I tried to talk him out of using if and using a separate server block > for 443 and 80 ports for each server, and just including common > configuration. ?Apparently, this does not currently work, but it seems > it should be more or less usable, I think he was having problems with > using default ssl in the listen line and NginX would not bind multiple > times. ?At any rate, he also complained about the "roundabout > redundancy" of the config. ?That got me to thinking, would something > like the following be desirable, and how difficult to implement would > it be? > > server { > ?listen 80; > ?listen 443; > ?ssl on 443; > ?... normal server config ... > } > > My gut instinct tells me this would be a lot more work than simply > allowing multiple SSL hosts per IP address, but it does seem to have a > nice ring to it, this late in the day. > > So to summarize, I am curious if 0.8.x is planned to support multiple > SSL hosts per IP, if that feature is even desired by anyone (or maybe > other ways to do it, in case I see this problem again!), and finally > anyone and everyone's thoughts on the above syntax to unify config for > SSL and non SSL (you could still use separate blocks if the > configuration is not exactly the same for plain and SSL, of course). > > Thanks, > Merlin > > From denis at gostats.ru Mon Jul 13 11:58:50 2009 From: denis at gostats.ru (Denis F. Latypoff) Date: Mon, 13 Jul 2009 14:58:50 +0700 Subject: SSL server_name support, single server block plain/SSL, et al in 0.8.x ? In-Reply-To: References: Message-ID: <186732783.20090713145850@gostats.ru> Hello merlin, Monday, July 13, 2009, 2:17:02 PM, you wrote: > Hello, > I had a long chat with a user in the IRC channel about SSL in nginx. > Initially, it started out talking about hosting multiple domains with > SSL on the same address. They pointed me to > http://tools.ietf.org/html/rfc4366#section-3.1 which seems fairly > straightforward and I am curious if there is interest in this or if > Igor plans to implement it at some point in future anyway. > Also, through talking with him, he showed me his idea for how the SSL > support should work. It went more or less like this: > ssl_certificate /etc/ssl/custom/supercatchall.crt; > ssl_certificate_key /etc/ssl/custom/supercatchall.key; > server { > listen 80; > listen 443; > if(port = 443) { > ssl on; > } > ... normal server config ... > } > ... repeated for several servers apparently on same IP ... > I tried to talk him out of using if and using a separate server block > for 443 and 80 ports for each server, and just including common > configuration. Apparently, this does not currently work, but it seems > it should be more or less usable, I think he was having problems with > using default ssl in the listen line and NginX would not bind multiple > times. At any rate, he also complained about the "roundabout > redundancy" of the config. That got me to thinking, would something > like the following be desirable, and how difficult to implement would > it be? > server { > listen 80; > listen 443; > ssl on 443; > ... normal server config ... > } > My gut instinct tells me this would be a lot more work than simply > allowing multiple SSL hosts per IP address, but it does seem to have a > nice ring to it, this late in the day. > So to summarize, I am curious if 0.8.x is planned to support multiple > SSL hosts per IP, if that feature is even desired by anyone (or maybe > other ways to do it, in case I see this problem again!), and finally > anyone and everyone's thoughts on the above syntax to unify config for > SSL and non SSL (you could still use separate blocks if the > configuration is not exactly the same for plain and SSL, of course). > Thanks, > Merlin http://wiki.nginx.org/NginxHttpCoreModule#listen server { listen 80; listen 443 ssl; ... normal server config ... } -- Best regards, Denis mailto:denis at gostats.ru From glen at lumanau.web.id Mon Jul 13 12:09:12 2009 From: glen at lumanau.web.id (Glen Lumanau) Date: Mon, 13 Jul 2009 15:09:12 +0700 Subject: [confused] a client request body is buffered to a temporary file Message-ID: <000001ca0391$366a2e30$a33e8a90$@web.id> I tried to find on google for the solution for this But still no luck 2009/07/13 15:05:22 [warn] 32467#0: *6899 a client request body is buffered to a temporary file /var/lib/nginx/body/0/00/0000000000, client: 192.168.254.16, server: mydomain.com, request: "POST /backend.php/F3/saveTV HTTP/1.0", host: "mydomain.com", referrer: http://mydomain.com/backend.php/F3/add Here's the config that I used server { listen 80 default; server_name mydomain.com www.mydomain.com; root /home/web; index index.php; charset utf-8; location ~ \.(gif|jpg|png|jpeg|swf|js|css|flv)$ { expires 30d; } location / { proxy_set_header X-Real-IP $remote_addr; proxy_set_header Host $http_host; proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for; proxy_pass http://127.0.0.1:8080; client_max_body_size 500m; client_body_buffer_size 1024k; proxy_connect_timeout 100; proxy_send_timeout 100; proxy_read_timeout 100; fastcgi_read_timeout 180; proxy_buffer_size 64k; proxy_buffers 4 64k; proxy_busy_buffers_size 128k; proxy_temp_file_write_size 128k; } } Help me please.. i'm so confused -------------- next part -------------- An HTML attachment was scrubbed... URL: From is at rambler-co.ru Mon Jul 13 12:07:21 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Mon, 13 Jul 2009 12:07:21 +0400 Subject: SSL server_name support, single server block plain/SSL, et al in 0.8.x ? In-Reply-To: References: Message-ID: <20090713080721.GA4512@rambler-co.ru> On Mon, Jul 13, 2009 at 12:17:02AM -0700, merlin corey wrote: > Hello, > > I had a long chat with a user in the IRC channel about SSL in nginx. > Initially, it started out talking about hosting multiple domains with > SSL on the same address. They pointed me to > http://tools.ietf.org/html/rfc4366#section-3.1 which seems fairly > straightforward and I am curious if there is interest in this or if > Igor plans to implement it at some point in future anyway. nginx supports SNI since 0.5.23. You just need OpenSSL built with SNI support. The main issue, however, are browsers: MSIE 6 and Windows XP do not support SNI and MSIE 6 is still in use: http://weblogs.mozillazine.org/asa/archives/2009/07/internet_explorer_6.html > Also, through talking with him, he showed me his idea for how the SSL > support should work. It went more or less like this: > > ssl_certificate /etc/ssl/custom/supercatchall.crt; > ssl_certificate_key /etc/ssl/custom/supercatchall.key; > server { > listen 80; > listen 443; > if(port = 443) { > ssl on; > } > ... normal server config ... > } > ... repeated for several servers apparently on same IP ... > > I tried to talk him out of using if and using a separate server block > for 443 and 80 ports for each server, and just including common > configuration. Apparently, this does not currently work, but it seems > it should be more or less usable, I think he was having problems with > using default ssl in the listen line and NginX would not bind multiple > times. At any rate, he also complained about the "roundabout > redundancy" of the config. That got me to thinking, would something > like the following be desirable, and how difficult to implement would > it be? > > server { > listen 80; > listen 443; > ssl on 443; > ... normal server config ... > } > > My gut instinct tells me this would be a lot more work than simply > allowing multiple SSL hosts per IP address, but it does seem to have a > nice ring to it, this late in the day. > > So to summarize, I am curious if 0.8.x is planned to support multiple > SSL hosts per IP, if that feature is even desired by anyone (or maybe > other ways to do it, in case I see this problem again!), and finally > anyone and everyone's thoughts on the above syntax to unify config for > SSL and non SSL (you could still use separate blocks if the > configuration is not exactly the same for plain and SSL, of course). Since 0.7.14 you can use: server { listen 80; listen 443 default ssl; ... -- Igor Sysoev http://sysoev.ru/en/ From is at rambler-co.ru Mon Jul 13 12:23:59 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Mon, 13 Jul 2009 12:23:59 +0400 Subject: [confused] a client request body is buffered to a temporary file In-Reply-To: <000001ca0391$366a2e30$a33e8a90$@web.id> References: <000001ca0391$366a2e30$a33e8a90$@web.id> Message-ID: <20090713082359.GB4512@rambler-co.ru> On Mon, Jul 13, 2009 at 03:09:12PM +0700, Glen Lumanau wrote: > I tried to find on google for the solution for this > > But still no luck > > > > 2009/07/13 15:05:22 [warn] 32467#0: *6899 a client request body is buffered > to a temporary file /var/lib/nginx/body/0/00/0000000000, client: > 192.168.254.16, server: mydomain.com, request: "POST /backend.php/F3/saveTV > HTTP/1.0", host: "mydomain.com", referrer: > http://mydomain.com/backend.php/F3/add > > > > Here's the config that I used > > > > server { > > listen 80 default; > > server_name mydomain.com www.mydomain.com; > > root /home/web; > > index index.php; > > > > charset utf-8; > > > > location ~ \.(gif|jpg|png|jpeg|swf|js|css|flv)$ { > > expires 30d; > > } > > > > location / { > > proxy_set_header X-Real-IP $remote_addr; > > proxy_set_header Host $http_host; > > proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for; > > > > proxy_pass http://127.0.0.1:8080; > > > > client_max_body_size 500m; > > client_body_buffer_size 1024k; You need to increase client_body_buffer_size. > proxy_connect_timeout 100; > > proxy_send_timeout 100; > > proxy_read_timeout 100; > > > > fastcgi_read_timeout 180; > > > > proxy_buffer_size 64k; > > proxy_buffers 4 64k; > > proxy_busy_buffers_size 128k; > > proxy_temp_file_write_size 128k; > > } > > } > > > > Help me please.. i'm so confused -- Igor Sysoev http://sysoev.ru/en/ From glen at lumanau.web.id Mon Jul 13 13:17:33 2009 From: glen at lumanau.web.id (Glen Lumanau) Date: Mon, 13 Jul 2009 16:17:33 +0700 Subject: [confused] a client request body is buffered to a temporary file In-Reply-To: <20090713082359.GB4512@rambler-co.ru> References: <000001ca0391$366a2e30$a33e8a90$@web.id> <20090713082359.GB4512@rambler-co.ru> Message-ID: <000501ca039a$c32550d0$496ff270$@web.id> Thank's it's working now But have onother issue I'm uploading a video to my backend. I saw no errors found on the error.log But the videos seems can't be copied to the destination folder. Before using nginx as reverse proxy, it can be copied without any problem. Any suggestions how can I check that? -----Original Message----- From: owner-nginx at sysoev.ru [mailto:owner-nginx at sysoev.ru] On Behalf Of Igor Sysoev Sent: 13 Juli 2009 15:24 To: nginx at sysoev.ru Subject: Re: [confused] a client request body is buffered to a temporary file On Mon, Jul 13, 2009 at 03:09:12PM +0700, Glen Lumanau wrote: > I tried to find on google for the solution for this > > But still no luck > > > > 2009/07/13 15:05:22 [warn] 32467#0: *6899 a client request body is buffered > to a temporary file /var/lib/nginx/body/0/00/0000000000, client: > 192.168.254.16, server: mydomain.com, request: "POST /backend.php/F3/saveTV > HTTP/1.0", host: "mydomain.com", referrer: > http://mydomain.com/backend.php/F3/add > > > > Here's the config that I used > > > > server { > > listen 80 default; > > server_name mydomain.com www.mydomain.com; > > root /home/web; > > index index.php; > > > > charset utf-8; > > > > location ~ \.(gif|jpg|png|jpeg|swf|js|css|flv)$ { > > expires 30d; > > } > > > > location / { > > proxy_set_header X-Real-IP $remote_addr; > > proxy_set_header Host $http_host; > > proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for; > > > > proxy_pass http://127.0.0.1:8080; > > > > client_max_body_size 500m; > > client_body_buffer_size 1024k; You need to increase client_body_buffer_size. > proxy_connect_timeout 100; > > proxy_send_timeout 100; > > proxy_read_timeout 100; > > > > fastcgi_read_timeout 180; > > > > proxy_buffer_size 64k; > > proxy_buffers 4 64k; > > proxy_busy_buffers_size 128k; > > proxy_temp_file_write_size 128k; > > } > > } > > > > Help me please.. i'm so confused -- Igor Sysoev http://sysoev.ru/en/ From mdounin at mdounin.ru Mon Jul 13 15:33:21 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Mon, 13 Jul 2009 15:33:21 +0400 Subject: Cache questions In-Reply-To: <4A57D142.2070501@gmail.com> References: <4A57D142.2070501@gmail.com> Message-ID: <20090713113321.GH67302@mdounin.ru> Hello! On Fri, Jul 10, 2009 at 07:39:46PM -0400, Jim Ohlstein wrote: > I'm using the fastcgi cache for static files (images, javascript,css) > and just found multiple lines in the error log like this one: > > 2009/07/10 10:22:54 [crit] 22476#0: ngx_slab_alloc() failed: no memory > in cache keys zone "one" > > So I increased the memory available for the zone and reloaded nginx. It > took over five hours to go through the cache but these are the relevant > entries: > > 2009/07/10 12:11:03 [notice] 21038#0: start cache manager process 32730 > 2009/07/10 12:11:04 [notice] 21038#0: cache manager process 22480 exited > with code 0 > > and finally > > 2009/07/10 17:43:27 [notice] 32730#0: http file cache: > /usr/local/nginx/cache 11638.289M, bsize: 4096 > > My questions are: > > Is that simply the total (11638.289MB or 11.4GB) of all of the file > sizes, or is that the actual disk space consumed taking into account > total number of blocks used multiplied by the block size? The number It's size on disk (i.e. number of blocks * block size), but for files only (it doesn't take directories into account). > could be quite different given what I estimate are nearly one million > mostly small files in the cache at this point. > > When I next upgrade nginx (I'm running 0.8.4), and I attempt a "graceful > upgrade" will it have to go through this entire process again? Yes. Binary upgrade doesn't preserve shared memory zones, so cache will be rescanned again. Note that nginx uses cache even before it was completely scanned, so it shouldn't be a problem. Maxim Dounin From mdounin at mdounin.ru Mon Jul 13 15:52:31 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Mon, 13 Jul 2009 15:52:31 +0400 Subject: Can't start nginx In-Reply-To: <808d0ac45d2eeeabbc7dbf3881810c7b@ruby-forum.com> References: <808d0ac45d2eeeabbc7dbf3881810c7b@ruby-forum.com> Message-ID: <20090713115231.GI67302@mdounin.ru> Hello! On Sat, Jul 11, 2009 at 05:13:20PM +0200, Andreas Vieten wrote: > Hello, > > I've tried a serveral times to start nginx but it always ends with no > happy end. > > Download and compiling is no problem for me. > > So here goes my way that i do it: > > 1) cd /tmp > wget wget http://sysoev.ru/nginx/nginx-0.7.61.tar.gz > tar xvfz nginx-0.7.61.tar.gz > cd nginx-0.7.61 > > ./configure \ > --prefix=/usr \ With this prefix and without --sbin-path set nginx will install it's binary into /usr/sbin/nginx. > --conf-path=/etc/nginx/nginx.conf \ > --http-log-path=/var/log/nginx/access_log \ > --error-log-path=/var/log/nginx/error_log \ > --pid-path=/var/run/nginx.pid \ > --http-client-body-temp-path=/var/tmp/nginx/client \ > --http-proxy-temp-path=/var/tmp/nginx/proxy \ > --http-fastcgi-temp-path=/var/tmp/nginx/fastcgi \ > --with-md5-asm --with-md5=/usr/include \ Do you have md5 library *sources* in /usr/include? Note that --with-md5 specifies path to md5 library sources (and nginx will try to build them), it's not path to headers. > --with-http_realip_module \ > --with-http_ssl_module \ > --with-http_stub_status_module > > make > make install > > 2) vi /etc/init.d/nginx > > > #! /bin/sh > > ### BEGIN INIT INFO > # Provides: nginx > # Required-Start: $all > # Required-Stop: $all > # Default-Start: 2 3 4 5 > # Default-Stop: 0 1 6 > # Short-Description: starts the nginx web server > # Description: starts nginx using start-stop-daemon > ### END INIT INFO > > PATH=/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin > DAEMON=/usr/local/sbin/nginx Seems to be using /usr/local/sbin/nginx binary, while you've just compiled /usr/sbin/nginx one (if build succeded - I'm not sure due to wierd configure arguments). Note that other parts of your init script assume /usr/local/... paths as well. [...] Maxim Dounin From is at rambler-co.ru Mon Jul 13 15:48:40 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Mon, 13 Jul 2009 15:48:40 +0400 Subject: Cache questions In-Reply-To: <20090713113321.GH67302@mdounin.ru> References: <4A57D142.2070501@gmail.com> <20090713113321.GH67302@mdounin.ru> Message-ID: <20090713114840.GE9584@rambler-co.ru> On Mon, Jul 13, 2009 at 03:33:21PM +0400, Maxim Dounin wrote: > Hello! > > On Fri, Jul 10, 2009 at 07:39:46PM -0400, Jim Ohlstein wrote: > > > I'm using the fastcgi cache for static files (images, javascript,css) > > and just found multiple lines in the error log like this one: > > > > 2009/07/10 10:22:54 [crit] 22476#0: ngx_slab_alloc() failed: no memory > > in cache keys zone "one" > > > > So I increased the memory available for the zone and reloaded nginx. It > > took over five hours to go through the cache but these are the relevant > > entries: > > > > 2009/07/10 12:11:03 [notice] 21038#0: start cache manager process 32730 > > 2009/07/10 12:11:04 [notice] 21038#0: cache manager process 22480 exited > > with code 0 > > > > and finally > > > > 2009/07/10 17:43:27 [notice] 32730#0: http file cache: > > /usr/local/nginx/cache 11638.289M, bsize: 4096 > > > > My questions are: > > > > Is that simply the total (11638.289MB or 11.4GB) of all of the file > > sizes, or is that the actual disk space consumed taking into account > > total number of blocks used multiplied by the block size? The number > > It's size on disk (i.e. number of blocks * block size), but for > files only (it doesn't take directories into account). Just note: nginx rounds a file size to the bsize. bsize is f_bsize from statfs() or f_frsize from statvfs(). I'm not sure that bsize matches always a filesystem allocation unit. > > could be quite different given what I estimate are nearly one million > > mostly small files in the cache at this point. > > > > When I next upgrade nginx (I'm running 0.8.4), and I attempt a "graceful > > upgrade" will it have to go through this entire process again? > > Yes. Binary upgrade doesn't preserve shared memory zones, so > cache will be rescanned again. > > Note that nginx uses cache even before it was completely scanned, > so it shouldn't be a problem. > > Maxim Dounin -- Igor Sysoev http://sysoev.ru/en/ From is at rambler-co.ru Mon Jul 13 15:54:53 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Mon, 13 Jul 2009 15:54:53 +0400 Subject: nginx-0.8.5 Message-ID: <20090713115453.GG9584@rambler-co.ru> Changes with nginx 0.8.5 13 Jul 2009 *) Bugfix: now nginx allows underscores in a request method. *) Bugfix: a 500 error code was returned for invalid login/password while HTTP Basic authentication on Windows. *) Bugfix: ngx_http_perl_module responses did not work in subrequests. *) Bugfix: in ngx_http_limit_req_module. Thanks to Maxim Dounin. -- Igor Sysoev http://sysoev.ru/en/ From lists at ruby-forum.com Mon Jul 13 16:14:29 2009 From: lists at ruby-forum.com (Robert Gabriel) Date: Mon, 13 Jul 2009 14:14:29 +0200 Subject: Can't start nginx In-Reply-To: <808d0ac45d2eeeabbc7dbf3881810c7b@ruby-forum.com> References: <808d0ac45d2eeeabbc7dbf3881810c7b@ruby-forum.com> Message-ID: <31f5a14539e362a7ad78e4cf31513f2b@ruby-forum.com> Try this: sh -x /etc/init.d/nginx start and see what happen or put it in a reply, as you wish. -- Posted via http://www.ruby-forum.com/. From mdounin at mdounin.ru Mon Jul 13 16:36:39 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Mon, 13 Jul 2009 16:36:39 +0400 Subject: behaviour changes upgrading from nginx .6.x to .7.x In-Reply-To: <4A58BD91.3090909@terabytemedia.com> References: <4A58BD91.3090909@terabytemedia.com> Message-ID: <20090713123639.GJ67302@mdounin.ru> Hello! On Sat, Jul 11, 2009 at 09:28:01AM -0700, Michael McCallister wrote: > Greetings, > > First, thanks for nginx - very cool web server. Next, upon upgrading > from 0.6.36 to 0.7.61 I noticed some changes in behaviour which prompted > me to downgrade back to 0.6.36 until I could further understand the > changes between .6.x and .7.x. More specifically, here is where things > worked differently: > > Given this rewrite rule: rewrite ^/[Aa]rticles/?(.*) /search.php?q=$1; > > This URL: /Articles/highway+diesel+vehicles > > Would generate the following $_GET in php (5.2.8 using fpm) and nginx > (0.6.36): > > array ( > 'q' => 'highway diesel vehicles', > ) > > Whereas the same exact URL would generate the following $_GET in php > (5.2.8 using fpm - no changes to php) and nginx (0.7.61 - nginx is the > only thing that changed): > > array ( > 'q' => 'highway+diesel+vehicles', > ) > > I view the behaviour under nginx 0.6.36 as correct while the behaviour > under 0.7.61 does not result in the expected value. Obviously it is > trivial to workaround this issue, but this leads up to my main concern - > what else behaves differently? I tried to find documentation outlining > upgrade considerations while migrating from .6.x to .7.x but could not > find any - my apologies if I missed them and/or am asking a question > already addressed previously on the list. > > So I guess I really have two questions considering the above: > > 1) Is the above behaviour in .7.x expected? Yes. Changes with 0.7.14: ... *) Bugfix: if URI part captured by a "rewrite" directive was used as a query string, then the query string was not escaped. ... Note that 'highway+diesel+vehicles' is *unescaped* value ('+' has no special meaning when used in path component), and it's then escaped when used in query string to 'highway%2bdiesel%2bvehicles' (since '+' reserved in query string). In it's turn php unescapes query string escaping and you see $_GET[q] = 'highway+diesel+vehicles' exactly as it was captured by your rewrite. > 2) Is there a list of things to watch out for when changing from .6.x to > .7.x? Changes are documented in CHANGES file in nginx sources (online version available at http://nginx.net/CHANGES). It's not exactly what you are asking about, but as far as I know it's all we currently have (and it worth reading anyway). Maxim Dounin > > Any help is greatly appreciated and thanks again for nginx. > > Michael > > From bing_ran at hotmail.com Mon Jul 13 18:38:58 2009 From: bing_ran at hotmail.com (=?gb2312?B?yL2x+A==?=) Date: Mon, 13 Jul 2009 22:38:58 +0800 Subject: HTTP limit Message-ID: Hi, I'm relatively new to nginx config scripting and I'm wondering if it's possible to exluding specific IP from either the limit_conn or the limit_req directive. Will enclosing the directive in a 'if' directive will do the trick and a sample? Thanks Bing -------------- next part -------------- An HTML attachment was scrubbed... URL: From jg at internetx.de Mon Jul 13 18:51:27 2009 From: jg at internetx.de (Juergen Gotteswinter) Date: Mon, 13 Jul 2009 16:51:27 +0200 Subject: nginx-0.8.5 In-Reply-To: <20090713115453.GG9584@rambler-co.ru> References: <20090713115453.GG9584@rambler-co.ru> Message-ID: <4A5B49EF.2030702@internetx.de> Hello Updated the SuSE RPMS in the opensuse-buildservice to 0.8.5 http://download.opensuse.org/repositories/server:/http/ Regards, On 07/13/2009 01:54 PM, Igor Sysoev wrote: > Changes with nginx 0.8.5 13 Jul 2009 > > *) Bugfix: now nginx allows underscores in a request method. > > *) Bugfix: a 500 error code was returned for invalid login/password > while HTTP Basic authentication on Windows. > > *) Bugfix: ngx_http_perl_module responses did not work in subrequests. > > *) Bugfix: in ngx_http_limit_req_module. > Thanks to Maxim Dounin. > > > -- InterNetX GmbH Maximilianstrasse 6 D-93047 Regensburg Tel. +49 941 59559-480 Fax +49 941 59559-245 Gesch?ftsf?hrer/CEO: Thomas M?rz Amtsgericht Regensburg, HRB 7142 From lists at ruby-forum.com Mon Jul 13 20:03:33 2009 From: lists at ruby-forum.com (Shai Rosenfeld) Date: Mon, 13 Jul 2009 18:03:33 +0200 Subject: Client-IP Rails issue In-Reply-To: References: Message-ID: Hope this helps someone - https://rails.lighthouseapp.com/projects/8994/tickets/322 Shai. -- Posted via http://www.ruby-forum.com/. From nginx-forum at nginx.us Mon Jul 13 20:31:01 2009 From: nginx-forum at nginx.us (AndiV74) Date: Mon, 13 Jul 2009 12:31:01 -0400 Subject: Can't start nginx In-Reply-To: <31f5a14539e362a7ad78e4cf31513f2b@forum.nginx.org> References: <808d0ac45d2eeeabbc7dbf3881810c7b@ruby-forum.com> <31f5a14539e362a7ad78e4cf31513f2b@forum.nginx.org> Message-ID: <00c4f2c0c5f5117ff008508449c76065.NginxMailingList@forum.nginx.org> Robert Gabriel Wrote: ------------------------------------------------------- > Try this: > > sh -x /etc/init.d/nginx start > > and see what happen or put it in a reply, as you > wish. > -- > Posted via http://www.ruby-forum.com/. Hello, it doesn't work either. Posted at Nginx Forum: http://forum.nginx.org/read.php?2,3989,4014#msg-4014 From support-nginx at oeko.net Mon Jul 13 21:04:09 2009 From: support-nginx at oeko.net (Toni Mueller) Date: Mon, 13 Jul 2009 19:04:09 +0200 Subject: 0.7.61: error_page 404 woes Message-ID: <20090713170409.23831.qmail@oak.oeko.net> Hi, I'm having a problem with the error_page directive. What I want is to redirect people who try to access a nonexistent page to the home page of the site. Following http://wiki.nginx.org/NginxHttpCoreModule#error_page , I configured error_page 404 /; But this statement had no effect. I had to re-write the statement as error_page 404 http://$server_name/; to get the desired effect, but don't quite understand, why. TIA! Kind regards, --Toni++ From mikemc-nginx at terabytemedia.com Mon Jul 13 21:48:30 2009 From: mikemc-nginx at terabytemedia.com (Michael McCallister) Date: Mon, 13 Jul 2009 10:48:30 -0700 Subject: behaviour changes upgrading from nginx .6.x to .7.x In-Reply-To: <20090713123639.GJ67302@mdounin.ru> References: <4A58BD91.3090909@terabytemedia.com> <20090713123639.GJ67302@mdounin.ru> Message-ID: <4A5B736E.80901@terabytemedia.com> Maxim Dounin wrote: > Hello! > > On Sat, Jul 11, 2009 at 09:28:01AM -0700, Michael McCallister wrote: > > Yes. Changes with 0.7.14: > > ... > > *) Bugfix: if URI part captured by a "rewrite" directive was used as a > query string, then the query string was not escaped. > > ... > > Note that 'highway+diesel+vehicles' is *unescaped* value ('+' > has no special meaning when used in path component), and it's > then escaped when used in query string to 'highway%2bdiesel%2bvehicles' > (since '+' reserved in query string). In it's turn php unescapes > query string escaping and you see > > $_GET[q] = 'highway+diesel+vehicles' > > exactly as it was captured by your rewrite. > > >> 2) Is there a list of things to watch out for when changing from .6.x to >> .7.x? >> > > Changes are documented in CHANGES file in nginx sources > (online version available at http://nginx.net/CHANGES). It's not > exactly what you are asking about, but as far as I know it's all > we currently have (and it worth reading anyway). > > Maxim Dounin Thanks From is at rambler-co.ru Mon Jul 13 22:22:25 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Mon, 13 Jul 2009 22:22:25 +0400 Subject: 0.7.61: error_page 404 woes In-Reply-To: <20090713170409.23831.qmail@oak.oeko.net> References: <20090713170409.23831.qmail@oak.oeko.net> Message-ID: <20090713182225.GB15437@rambler-co.ru> On Mon, Jul 13, 2009 at 07:04:09PM +0200, Toni Mueller wrote: > > > Hi, > > I'm having a problem with the error_page directive. > > What I want is to redirect people who try to access a nonexistent page > to the home page of the site. Following > http://wiki.nginx.org/NginxHttpCoreModule#error_page , I configured > > error_page 404 /; > > But this statement had no effect. I had to re-write the statement as > > error_page 404 http://$server_name/; > > to get the desired effect, but don't quite understand, why. Because the first statement does an internal redirect to /, while the second one does an external redirect (302). -- Igor Sysoev http://sysoev.ru/en/ From support at oeko.net Mon Jul 13 22:48:29 2009 From: support at oeko.net (Toni Mueller) Date: Mon, 13 Jul 2009 20:48:29 +0200 Subject: 0.7.61: error_page 404 woes In-Reply-To: <20090713182225.GB15437@rambler-co.ru> References: <20090713170409.23831.qmail@oak.oeko.net> <20090713182225.GB15437@rambler-co.ru> Message-ID: <20090713184829.27178.qmail@oak.oeko.net> Hi, On Mon, 13.07.2009 at 22:22:25 +0400, Igor Sysoev wrote: > On Mon, Jul 13, 2009 at 07:04:09PM +0200, Toni Mueller wrote: > > error_page 404 /; > > > > But this statement had no effect. I had to re-write the statement as > > > > error_page 404 http://$server_name/; > > > > to get the desired effect, but don't quite understand, why. > > Because the first statement does an internal redirect to /, > while the second one does an external redirect (302). thanks for the quick answer, but I don't quite understand: I already assumed that the second would cause an external redirect, but why the first statement would not lead to an internal rewrite of the URI and then try that through the remainder of the configuration, I don't see. I mean, the documentation suggests that "error_page 404 /somepage.html" should work, and I don't see that much of a difference. Kind regards, --Toni++ From is at rambler-co.ru Mon Jul 13 23:57:34 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Mon, 13 Jul 2009 23:57:34 +0400 Subject: 0.7.61: error_page 404 woes In-Reply-To: <20090713184829.27178.qmail@oak.oeko.net> References: <20090713170409.23831.qmail@oak.oeko.net> <20090713182225.GB15437@rambler-co.ru> <20090713184829.27178.qmail@oak.oeko.net> Message-ID: <20090713195734.GD15437@rambler-co.ru> On Mon, Jul 13, 2009 at 08:48:29PM +0200, Toni Mueller wrote: > > Hi, > > On Mon, 13.07.2009 at 22:22:25 +0400, Igor Sysoev wrote: > > On Mon, Jul 13, 2009 at 07:04:09PM +0200, Toni Mueller wrote: > > > error_page 404 /; > > > > > > But this statement had no effect. I had to re-write the statement as > > > > > > error_page 404 http://$server_name/; > > > > > > to get the desired effect, but don't quite understand, why. > > > > Because the first statement does an internal redirect to /, > > while the second one does an external redirect (302). > > thanks for the quick answer, but I don't quite understand: > > I already assumed that the second would cause an external redirect, but > why the first statement would not lead to an internal rewrite of the > URI and then try that through the remainder of the configuration, I > don't see. What do you see in the first case ? You should see main page with unchanged nonexistent URI in browser. > I mean, the documentation suggests that "error_page 404 /somepage.html" > should work, and I don't see that much of a difference. -- Igor Sysoev http://sysoev.ru/en/ From jim.ohlstein at gmail.com Tue Jul 14 00:02:06 2009 From: jim.ohlstein at gmail.com (Jim Ohlstein) Date: Mon, 13 Jul 2009 16:02:06 -0400 Subject: Cache questions In-Reply-To: <20090713114840.GE9584@rambler-co.ru> References: <4A57D142.2070501@gmail.com> <20090713113321.GH67302@mdounin.ru> <20090713114840.GE9584@rambler-co.ru> Message-ID: <4A5B92BE.60002@gmail.com> Igor Sysoev wrote: > On Mon, Jul 13, 2009 at 03:33:21PM +0400, Maxim Dounin wrote: > > >> Hello! >> >> On Fri, Jul 10, 2009 at 07:39:46PM -0400, Jim Ohlstein wrote: >> >> >>> I'm using the fastcgi cache for static files (images, javascript,css) >>> and just found multiple lines in the error log like this one: >>> >>> 2009/07/10 10:22:54 [crit] 22476#0: ngx_slab_alloc() failed: no memory >>> in cache keys zone "one" >>> >>> So I increased the memory available for the zone and reloaded nginx. It >>> took over five hours to go through the cache but these are the relevant >>> entries: >>> >>> 2009/07/10 12:11:03 [notice] 21038#0: start cache manager process 32730 >>> 2009/07/10 12:11:04 [notice] 21038#0: cache manager process 22480 exited >>> with code 0 >>> >>> and finally >>> >>> 2009/07/10 17:43:27 [notice] 32730#0: http file cache: >>> /usr/local/nginx/cache 11638.289M, bsize: 4096 >>> >>> My questions are: >>> >>> Is that simply the total (11638.289MB or 11.4GB) of all of the file >>> sizes, or is that the actual disk space consumed taking into account >>> total number of blocks used multiplied by the block size? The number >>> >> It's size on disk (i.e. number of blocks * block size), but for >> files only (it doesn't take directories into account). >> > > Just note: nginx rounds a file size to the bsize. > bsize is f_bsize from statfs() or f_frsize from statvfs(). > I'm not sure that bsize matches always a filesystem allocation unit. > OK, Thanks. In trying to tune this, if I set fastcgi_cache_min_uses to 2, does that mean that a file will only be written to the cache the second time that it is requested? Google translate did not give me a clear answer to this from the Russian documentation. I think that I could improve efficiency greatly if I didn't cache files on the first request. I would know better if I could get some statistics. I know the last time I asked the answer was "not yet". Do you have any idea when this might be implemented even on a rudimentary basis? > >>> could be quite different given what I estimate are nearly one million >>> mostly small files in the cache at this point. >>> >>> When I next upgrade nginx (I'm running 0.8.4), and I attempt a "graceful >>> upgrade" will it have to go through this entire process again? >>> >> Yes. Binary upgrade doesn't preserve shared memory zones, so >> cache will be rescanned again. >> >> Note that nginx uses cache even before it was completely scanned, >> so it shouldn't be a problem. >> Yes, but it uses a fair bit of system resources scanning a large cache. Nothing much to be done about it though. >> Maxim Dounin >> > > > From is at rambler-co.ru Tue Jul 14 00:17:46 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Tue, 14 Jul 2009 00:17:46 +0400 Subject: Cache questions In-Reply-To: <4A5B92BE.60002@gmail.com> References: <4A57D142.2070501@gmail.com> <20090713113321.GH67302@mdounin.ru> <20090713114840.GE9584@rambler-co.ru> <4A5B92BE.60002@gmail.com> Message-ID: <20090713201746.GE15437@rambler-co.ru> On Mon, Jul 13, 2009 at 04:02:06PM -0400, Jim Ohlstein wrote: > > > Igor Sysoev wrote: > >On Mon, Jul 13, 2009 at 03:33:21PM +0400, Maxim Dounin wrote: > > > > > >>Hello! > >> > >>On Fri, Jul 10, 2009 at 07:39:46PM -0400, Jim Ohlstein wrote: > >> > >> > >>>I'm using the fastcgi cache for static files (images, javascript,css) > >>>and just found multiple lines in the error log like this one: > >>> > >>>2009/07/10 10:22:54 [crit] 22476#0: ngx_slab_alloc() failed: no memory > >>>in cache keys zone "one" > >>> > >>>So I increased the memory available for the zone and reloaded nginx. It > >>>took over five hours to go through the cache but these are the relevant > >>>entries: > >>> > >>>2009/07/10 12:11:03 [notice] 21038#0: start cache manager process 32730 > >>>2009/07/10 12:11:04 [notice] 21038#0: cache manager process 22480 exited > >>>with code 0 > >>> > >>>and finally > >>> > >>>2009/07/10 17:43:27 [notice] 32730#0: http file cache: > >>>/usr/local/nginx/cache 11638.289M, bsize: 4096 > >>> > >>>My questions are: > >>> > >>>Is that simply the total (11638.289MB or 11.4GB) of all of the file > >>>sizes, or is that the actual disk space consumed taking into account > >>>total number of blocks used multiplied by the block size? The number > >>> > >>It's size on disk (i.e. number of blocks * block size), but for > >>files only (it doesn't take directories into account). > >> > > > >Just note: nginx rounds a file size to the bsize. > >bsize is f_bsize from statfs() or f_frsize from statvfs(). > >I'm not sure that bsize matches always a filesystem allocation unit. > > > > OK, Thanks. > > In trying to tune this, if I set fastcgi_cache_min_uses to 2, does that > mean that a file will only be written to the cache the second time that > it is requested? Google translate did not give me a clear answer to this > from the Russian documentation. I think that I could improve efficiency > greatly if I didn't cache files on the first request. Yes, a file will be cached only on the second request made during period set in by "inactive" option of fastcgi_cache_path directive. > I would know > better if I could get some statistics. I know the last time I asked the > answer was "not yet". Do you have any idea when this might be > implemented even on a rudimentary basis? You may log $upstream_cache_status. > >>>could be quite different given what I estimate are nearly one million > >>>mostly small files in the cache at this point. > >>> > >>>When I next upgrade nginx (I'm running 0.8.4), and I attempt a "graceful > >>>upgrade" will it have to go through this entire process again? > >>> > >>Yes. Binary upgrade doesn't preserve shared memory zones, so > >>cache will be rescanned again. > >> > >>Note that nginx uses cache even before it was completely scanned, > >>so it shouldn't be a problem. > >> > > Yes, but it uses a fair bit of system resources scanning a large cache. > Nothing much to be done about it though. I tried to minimize impact of cache scanning according to this comment: /* * if processing 100 files takes more than 200ms, * it seems that many operations require disk i/o, * therefore sleep 200ms */ -- Igor Sysoev http://sysoev.ru/en/ From mdounin at mdounin.ru Tue Jul 14 00:50:54 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Tue, 14 Jul 2009 00:50:54 +0400 Subject: Cache questions In-Reply-To: <4A5B92BE.60002@gmail.com> References: <4A57D142.2070501@gmail.com> <20090713113321.GH67302@mdounin.ru> <20090713114840.GE9584@rambler-co.ru> <4A5B92BE.60002@gmail.com> Message-ID: <20090713205054.GP67302@mdounin.ru> Hello! On Mon, Jul 13, 2009 at 04:02:06PM -0400, Jim Ohlstein wrote: > > > Igor Sysoev wrote: >> On Mon, Jul 13, 2009 at 03:33:21PM +0400, Maxim Dounin wrote: >> >> >>> Hello! >>> >>> On Fri, Jul 10, 2009 at 07:39:46PM -0400, Jim Ohlstein wrote: >>> >>> >>>> I'm using the fastcgi cache for static files (images, >>>> javascript,css) and just found multiple lines in the error log >>>> like this one: >>>> >>>> 2009/07/10 10:22:54 [crit] 22476#0: ngx_slab_alloc() failed: no >>>> memory in cache keys zone "one" >>>> >>>> So I increased the memory available for the zone and reloaded >>>> nginx. It took over five hours to go through the cache but these >>>> are the relevant entries: >>>> >>>> 2009/07/10 12:11:03 [notice] 21038#0: start cache manager process 32730 >>>> 2009/07/10 12:11:04 [notice] 21038#0: cache manager process 22480 >>>> exited with code 0 >>>> >>>> and finally >>>> >>>> 2009/07/10 17:43:27 [notice] 32730#0: http file cache: >>>> /usr/local/nginx/cache 11638.289M, bsize: 4096 >>>> >>>> My questions are: >>>> >>>> Is that simply the total (11638.289MB or 11.4GB) of all of the file >>>> sizes, or is that the actual disk space consumed taking into >>>> account total number of blocks used multiplied by the block size? >>>> The number >>> It's size on disk (i.e. number of blocks * block size), but for >>> files only (it doesn't take directories into account). >>> >> >> Just note: nginx rounds a file size to the bsize. >> bsize is f_bsize from statfs() or f_frsize from statvfs(). >> I'm not sure that bsize matches always a filesystem allocation unit. >> > > OK, Thanks. > > In trying to tune this, if I set fastcgi_cache_min_uses to 2, does that > mean that a file will only be written to the cache the second time that > it is requested? Google translate did not give me a clear answer to this > from the Russian documentation. I think that I could improve efficiency > greatly if I didn't cache files on the first request. Yes, file will be written only after second request. Note that shared memory will be allocated anyway. > I would know > better if I could get some statistics. I know the last time I asked the > answer was "not yet". Do you have any idea when this might be > implemented even on a rudimentary basis? Cache status was implemented in 0.8.3: *) Feature: the $upstream_cache_status variable. You may log it to access_log and collect stats you want. >>>> could be quite different given what I estimate are nearly one >>>> million mostly small files in the cache at this point. >>>> >>>> When I next upgrade nginx (I'm running 0.8.4), and I attempt a >>>> "graceful upgrade" will it have to go through this entire process >>>> again? >>>> >>> Yes. Binary upgrade doesn't preserve shared memory zones, so cache >>> will be rescanned again. >>> >>> Note that nginx uses cache even before it was completely scanned, so >>> it shouldn't be a problem. >>> > > Yes, but it uses a fair bit of system resources scanning a large cache. > Nothing much to be done about it though. Actually, nginx tries to be nice and pauses scanning for 200ms if takes more than 200ms to scan 100 files. Maxim Dounin From lists at ruby-forum.com Tue Jul 14 01:23:36 2009 From: lists at ruby-forum.com (Andreas Vieten) Date: Mon, 13 Jul 2009 23:23:36 +0200 Subject: Can't start nginx In-Reply-To: <31f5a14539e362a7ad78e4cf31513f2b@ruby-forum.com> References: <808d0ac45d2eeeabbc7dbf3881810c7b@ruby-forum.com> <31f5a14539e362a7ad78e4cf31513f2b@ruby-forum.com> Message-ID: <6ede0e531f82cbd739f53b739d7971a6@ruby-forum.com> Robert Gabriel wrote: > Try this: > > sh -x /etc/init.d/nginx start > > and see what happen or put it in a reply, as you wish. Hello, doesn't work either. -- Posted via http://www.ruby-forum.com/. From merlincorey at dc949.org Tue Jul 14 04:41:12 2009 From: merlincorey at dc949.org (merlin corey) Date: Mon, 13 Jul 2009 17:41:12 -0700 Subject: SSL server_name support, single server block plain/SSL, et al in 0.8.x ? In-Reply-To: <20090713080721.GA4512@rambler-co.ru> References: <20090713080721.GA4512@rambler-co.ru> Message-ID: 2009/7/13 Igor Sysoev : > On Mon, Jul 13, 2009 at 12:17:02AM -0700, merlin corey wrote: > >> Hello, >> >> I had a long chat with a user in the IRC channel about SSL in nginx. >> Initially, it started out talking about hosting multiple domains with >> SSL on the same address. ?They pointed me to >> http://tools.ietf.org/html/rfc4366#section-3.1 which seems fairly >> straightforward and I am curious if there is interest in this or if >> Igor plans to implement it at some point in future anyway. > > nginx supports SNI since 0.5.23. You just need OpenSSL built with > SNI support. The main issue, however, are browsers: MSIE 6 and Windows XP > do not support SNI and MSIE 6 is still in use: > http://weblogs.mozillazine.org/asa/archives/2009/07/internet_explorer_6.html > >> Also, through talking with him, he showed me his idea for how the SSL >> support should work. ?It went more or less like this: >> >> ssl_certificate ? ? ?/etc/ssl/custom/supercatchall.crt; >> ssl_certificate_key ?/etc/ssl/custom/supercatchall.key; >> server { >> ? listen 80; >> ? listen 443; >> ? if(port = 443) { >> ? ? ssl on; >> ? } >> ? ... normal server config ... >> } >> ... repeated for several servers apparently on same IP ... >> >> I tried to talk him out of using if and using a separate server block >> for 443 and 80 ports for each server, and just including common >> configuration. ?Apparently, this does not currently work, but it seems >> it should be more or less usable, I think he was having problems with >> using default ssl in the listen line and NginX would not bind multiple >> times. ?At any rate, he also complained about the "roundabout >> redundancy" of the config. ?That got me to thinking, would something >> like the following be desirable, and how difficult to implement would >> it be? >> >> server { >> ? listen 80; >> ? listen 443; >> ? ssl on 443; >> ? ... normal server config ... >> } >> >> My gut instinct tells me this would be a lot more work than simply >> allowing multiple SSL hosts per IP address, but it does seem to have a >> nice ring to it, this late in the day. >> >> So to summarize, I am curious if 0.8.x is planned to support multiple >> SSL hosts per IP, if that feature is even desired by anyone (or maybe >> other ways to do it, in case I see this problem again!), and finally >> anyone and everyone's thoughts on the above syntax to unify config for >> SSL and non SSL (you could still use separate blocks if the >> configuration is not exactly the same for plain and SSL, of course). > Since 0.7.14 you can use: > > ? ?server { > ? ? ? listen 80; > ? ? ? listen 443 default ssl; > ? ? ? ... > Ah perfect, I was ignorant of this change and I see it is already on the English wiki so I have no excuse really for missing it. Goodbye ssl on blah blah; I knew there had to be an easier way :D. Thanks, Igor and Denis. Almir, I'm very aware that you don't need ifs, thank you; I head the get rid of if's in nginx configurations campaign on IRC :D. As to this multiple SSL domains on a single IP thing... I guess I may as well describe what he was trying to do. I am not sure if it fits into SNI, but I will check it out and tell him about it as well. Basically, he wants to have a bunch of different subdomains on the same IP to have their own SSL certs and for some reason he does not want to use wildcard certificates or different ports or anything... he wants to just have it work, and it doesn't allow it. I'm not quite sure what to tell him, aside from what I already have. Thanks again, Merlin From jim.ohlstein at gmail.com Tue Jul 14 04:45:03 2009 From: jim.ohlstein at gmail.com (Jim Ohlstein) Date: Mon, 13 Jul 2009 20:45:03 -0400 Subject: Cache questions In-Reply-To: <20090713201746.GE15437@rambler-co.ru> References: <4A57D142.2070501@gmail.com> <20090713113321.GH67302@mdounin.ru> <20090713114840.GE9584@rambler-co.ru> <4A5B92BE.60002@gmail.com> <20090713201746.GE15437@rambler-co.ru> Message-ID: <4A5BD50F.1020705@gmail.com> Igor Sysoev wrote: > On Mon, Jul 13, 2009 at 04:02:06PM -0400, Jim Ohlstein wrote: > > >> Igor Sysoev wrote: >> >>> On Mon, Jul 13, 2009 at 03:33:21PM +0400, Maxim Dounin wrote: >>> >>> >>> >>>> Hello! >>>> >>>> On Fri, Jul 10, 2009 at 07:39:46PM -0400, Jim Ohlstein wrote: >>>> >>>> >>>> >>>>> I'm using the fastcgi cache for static files (images, javascript,css) >>>>> and just found multiple lines in the error log like this one: >>>>> >>>>> 2009/07/10 10:22:54 [crit] 22476#0: ngx_slab_alloc() failed: no memory >>>>> in cache keys zone "one" >>>>> >>>>> So I increased the memory available for the zone and reloaded nginx. It >>>>> took over five hours to go through the cache but these are the relevant >>>>> entries: >>>>> >>>>> 2009/07/10 12:11:03 [notice] 21038#0: start cache manager process 32730 >>>>> 2009/07/10 12:11:04 [notice] 21038#0: cache manager process 22480 exited >>>>> with code 0 >>>>> >>>>> and finally >>>>> >>>>> 2009/07/10 17:43:27 [notice] 32730#0: http file cache: >>>>> /usr/local/nginx/cache 11638.289M, bsize: 4096 >>>>> >>>>> My questions are: >>>>> >>>>> Is that simply the total (11638.289MB or 11.4GB) of all of the file >>>>> sizes, or is that the actual disk space consumed taking into account >>>>> total number of blocks used multiplied by the block size? The number >>>>> >>>>> >>>> It's size on disk (i.e. number of blocks * block size), but for >>>> files only (it doesn't take directories into account). >>>> >>>> >>> Just note: nginx rounds a file size to the bsize. >>> bsize is f_bsize from statfs() or f_frsize from statvfs(). >>> I'm not sure that bsize matches always a filesystem allocation unit. >>> >>> >> OK, Thanks. >> >> In trying to tune this, if I set fastcgi_cache_min_uses to 2, does that >> mean that a file will only be written to the cache the second time that >> it is requested? Google translate did not give me a clear answer to this >> from the Russian documentation. I think that I could improve efficiency >> greatly if I didn't cache files on the first request. >> > > Yes, a file will be cached only on the second request made during period > set in by "inactive" option of fastcgi_cache_path directive. > > >> I would know >> better if I could get some statistics. I know the last time I asked the >> answer was "not yet". Do you have any idea when this might be >> implemented even on a rudimentary basis? >> > > You may log $upstream_cache_status. > Can you give me an example of how to do this? I have tried various permutations of: location ~ (jpg|jpeg|png|gif|ico|js|css)$ { fastcgi_pass unix:/tmp/my.sock; fastcgi_cache one; fastcgi_cache_key unix:/tmp/my.sock.1$request_uri; fastcgi_ignore_headers Cache-Control Expires; fastcgi_cache_valid 200 302 1d; fastcgi_cache_valid 301 7d; fastcgi_cache_valid any 10m; fastcgi_cache_min_uses 2; include /usr/local/nginx/conf/fastcgi_params; fastcgi_param SCRIPT_NAME myscript; fastcgi_buffers 64 8k; access_log logs/my-cache.access.log; log_format '$remote_addr - $remote_user [$time_local] ' '"$request" $status $body_bytes_sent ' '"$http_referer" "$http_user_agent"' ' $upstream_cache_status '; } The logs give me all above with exception of "$upstream_cache_status". Every entry ends with the user agent. > >>>>> could be quite different given what I estimate are nearly one million >>>>> mostly small files in the cache at this point. >>>>> >>>>> When I next upgrade nginx (I'm running 0.8.4), and I attempt a "graceful >>>>> upgrade" will it have to go through this entire process again? >>>>> >>>>> >>>> Yes. Binary upgrade doesn't preserve shared memory zones, so >>>> cache will be rescanned again. >>>> >>>> Note that nginx uses cache even before it was completely scanned, >>>> so it shouldn't be a problem. >>>> >>>> >> Yes, but it uses a fair bit of system resources scanning a large cache. >> Nothing much to be done about it though. >> > > I tried to minimize impact of cache scanning according to this comment: > > /* > * if processing 100 files takes more than 200ms, > * it seems that many operations require disk i/o, > * therefore sleep 200ms > */ > > > Jim From mdounin at mdounin.ru Tue Jul 14 06:29:59 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Tue, 14 Jul 2009 06:29:59 +0400 Subject: Cache questions In-Reply-To: <4A5BD50F.1020705@gmail.com> References: <4A57D142.2070501@gmail.com> <20090713113321.GH67302@mdounin.ru> <20090713114840.GE9584@rambler-co.ru> <4A5B92BE.60002@gmail.com> <20090713201746.GE15437@rambler-co.ru> <4A5BD50F.1020705@gmail.com> Message-ID: <20090714022959.GR67302@mdounin.ru> Hello! On Mon, Jul 13, 2009 at 08:45:03PM -0400, Jim Ohlstein wrote: > > > Igor Sysoev wrote: >> On Mon, Jul 13, 2009 at 04:02:06PM -0400, Jim Ohlstein wrote: >> >> >>> Igor Sysoev wrote: >>> >>>> On Mon, Jul 13, 2009 at 03:33:21PM +0400, Maxim Dounin wrote: >>>> >>>> >>>>> Hello! >>>>> >>>>> On Fri, Jul 10, 2009 at 07:39:46PM -0400, Jim Ohlstein wrote: >>>>> >>>>> >>>>>> I'm using the fastcgi cache for static files (images, >>>>>> javascript,css) and just found multiple lines in the error log >>>>>> like this one: >>>>>> >>>>>> 2009/07/10 10:22:54 [crit] 22476#0: ngx_slab_alloc() failed: no >>>>>> memory in cache keys zone "one" >>>>>> >>>>>> So I increased the memory available for the zone and reloaded >>>>>> nginx. It took over five hours to go through the cache but >>>>>> these are the relevant entries: >>>>>> >>>>>> 2009/07/10 12:11:03 [notice] 21038#0: start cache manager process 32730 >>>>>> 2009/07/10 12:11:04 [notice] 21038#0: cache manager process >>>>>> 22480 exited with code 0 >>>>>> >>>>>> and finally >>>>>> >>>>>> 2009/07/10 17:43:27 [notice] 32730#0: http file cache: >>>>>> /usr/local/nginx/cache 11638.289M, bsize: 4096 >>>>>> >>>>>> My questions are: >>>>>> >>>>>> Is that simply the total (11638.289MB or 11.4GB) of all of the >>>>>> file sizes, or is that the actual disk space consumed taking >>>>>> into account total number of blocks used multiplied by the >>>>>> block size? The number >>>>> It's size on disk (i.e. number of blocks * block size), but for >>>>> files only (it doesn't take directories into account). >>>>> >>>> Just note: nginx rounds a file size to the bsize. >>>> bsize is f_bsize from statfs() or f_frsize from statvfs(). >>>> I'm not sure that bsize matches always a filesystem allocation unit. >>>> >>> OK, Thanks. >>> >>> In trying to tune this, if I set fastcgi_cache_min_uses to 2, does >>> that mean that a file will only be written to the cache the second >>> time that it is requested? Google translate did not give me a clear >>> answer to this from the Russian documentation. I think that I could >>> improve efficiency greatly if I didn't cache files on the first >>> request. >>> >> >> Yes, a file will be cached only on the second request made during period >> set in by "inactive" option of fastcgi_cache_path directive. >> >> >>> I would know better if I could get some statistics. I know the last >>> time I asked the answer was "not yet". Do you have any idea when this >>> might be implemented even on a rudimentary basis? >>> >> >> You may log $upstream_cache_status. >> > > Can you give me an example of how to do this? > > I have tried various permutations of: > > location ~ (jpg|jpeg|png|gif|ico|js|css)$ { > fastcgi_pass unix:/tmp/my.sock; > fastcgi_cache one; > fastcgi_cache_key unix:/tmp/my.sock.1$request_uri; > fastcgi_ignore_headers Cache-Control Expires; > fastcgi_cache_valid 200 302 1d; > fastcgi_cache_valid 301 7d; > fastcgi_cache_valid any 10m; > fastcgi_cache_min_uses 2; > include /usr/local/nginx/conf/fastcgi_params; > fastcgi_param SCRIPT_NAME myscript; > fastcgi_buffers 64 8k; > access_log logs/my-cache.access.log; > log_format '$remote_addr - $remote_user [$time_local] ' > '"$request" $status $body_bytes_sent ' > '"$http_referer" "$http_user_agent"' ' > $upstream_cache_status '; > } > > The logs give me all above with exception of "$upstream_cache_status". > Every entry ends with the user agent. log_format cachestatus '$remote_addr - $remote_user [$time_local] ' '"$request" $status $body_bytes_sent ' '"$http_referer" "$http_user_agent" ' '$upstream_cache_status'; access_log /path/to/log cachestatus; Note that log_format's first argument is format name, and second argument of access_log is format name too (optional, defaults to 'combined' predefined format). See http://wiki.nginx.org/NginxHttpLogModule for details. Maxim Dounin From jim.ohlstein at gmail.com Tue Jul 14 07:09:34 2009 From: jim.ohlstein at gmail.com (Jim Ohlstein) Date: Mon, 13 Jul 2009 23:09:34 -0400 Subject: Cache questions In-Reply-To: <20090714022959.GR67302@mdounin.ru> References: <4A57D142.2070501@gmail.com> <20090713113321.GH67302@mdounin.ru> <20090713114840.GE9584@rambler-co.ru> <4A5B92BE.60002@gmail.com> <20090713201746.GE15437@rambler-co.ru> <4A5BD50F.1020705@gmail.com> <20090714022959.GR67302@mdounin.ru> Message-ID: <4A5BF6EE.4080306@gmail.com> Maxim Dounin wrote: > Hello! > > On Mon, Jul 13, 2009 at 08:45:03PM -0400, Jim Ohlstein wrote: > > >> Igor Sysoev wrote: >> >>> On Mon, Jul 13, 2009 at 04:02:06PM -0400, Jim Ohlstein wrote: >>> >>> >>> >>>> Igor Sysoev wrote: >>>> >>>> >>>>> On Mon, Jul 13, 2009 at 03:33:21PM +0400, Maxim Dounin wrote: >>>>> >>>>> >>>>> >>>>>> Hello! >>>>>> >>>>>> On Fri, Jul 10, 2009 at 07:39:46PM -0400, Jim Ohlstein wrote: >>>>>> >>>>>> >>>>>> >>>>>>> I'm using the fastcgi cache for static files (images, >>>>>>> javascript,css) and just found multiple lines in the error log >>>>>>> like this one: >>>>>>> >>>>>>> 2009/07/10 10:22:54 [crit] 22476#0: ngx_slab_alloc() failed: no >>>>>>> memory in cache keys zone "one" >>>>>>> >>>>>>> So I increased the memory available for the zone and reloaded >>>>>>> nginx. It took over five hours to go through the cache but >>>>>>> these are the relevant entries: >>>>>>> >>>>>>> 2009/07/10 12:11:03 [notice] 21038#0: start cache manager process 32730 >>>>>>> 2009/07/10 12:11:04 [notice] 21038#0: cache manager process >>>>>>> 22480 exited with code 0 >>>>>>> >>>>>>> and finally >>>>>>> >>>>>>> 2009/07/10 17:43:27 [notice] 32730#0: http file cache: >>>>>>> /usr/local/nginx/cache 11638.289M, bsize: 4096 >>>>>>> >>>>>>> My questions are: >>>>>>> >>>>>>> Is that simply the total (11638.289MB or 11.4GB) of all of the >>>>>>> file sizes, or is that the actual disk space consumed taking >>>>>>> into account total number of blocks used multiplied by the >>>>>>> block size? The number >>>>>>> >>>>>> It's size on disk (i.e. number of blocks * block size), but for >>>>>> files only (it doesn't take directories into account). >>>>>> >>>>>> >>>>> Just note: nginx rounds a file size to the bsize. >>>>> bsize is f_bsize from statfs() or f_frsize from statvfs(). >>>>> I'm not sure that bsize matches always a filesystem allocation unit. >>>>> >>>>> >>>> OK, Thanks. >>>> >>>> In trying to tune this, if I set fastcgi_cache_min_uses to 2, does >>>> that mean that a file will only be written to the cache the second >>>> time that it is requested? Google translate did not give me a clear >>>> answer to this from the Russian documentation. I think that I could >>>> improve efficiency greatly if I didn't cache files on the first >>>> request. >>>> >>>> >>> Yes, a file will be cached only on the second request made during period >>> set in by "inactive" option of fastcgi_cache_path directive. >>> >>> >>> >>>> I would know better if I could get some statistics. I know the last >>>> time I asked the answer was "not yet". Do you have any idea when this >>>> might be implemented even on a rudimentary basis? >>>> >>>> >>> You may log $upstream_cache_status. >>> >>> >> Can you give me an example of how to do this? >> >> I have tried various permutations of: >> >> location ~ (jpg|jpeg|png|gif|ico|js|css)$ { >> fastcgi_pass unix:/tmp/my.sock; >> fastcgi_cache one; >> fastcgi_cache_key unix:/tmp/my.sock.1$request_uri; >> fastcgi_ignore_headers Cache-Control Expires; >> fastcgi_cache_valid 200 302 1d; >> fastcgi_cache_valid 301 7d; >> fastcgi_cache_valid any 10m; >> fastcgi_cache_min_uses 2; >> include /usr/local/nginx/conf/fastcgi_params; >> fastcgi_param SCRIPT_NAME myscript; >> fastcgi_buffers 64 8k; >> access_log logs/my-cache.access.log; >> log_format '$remote_addr - $remote_user [$time_local] ' >> '"$request" $status $body_bytes_sent ' >> '"$http_referer" "$http_user_agent"' ' >> $upstream_cache_status '; >> } >> >> The logs give me all above with exception of "$upstream_cache_status". >> Every entry ends with the user agent. >> > > log_format cachestatus '$remote_addr - $remote_user [$time_local] ' > '"$request" $status $body_bytes_sent ' > '"$http_referer" "$http_user_agent" ' > '$upstream_cache_status'; > > access_log /path/to/log cachestatus; > > Note that log_format's first argument is format name, and second > argument of access_log is format name too (optional, defaults > to 'combined' predefined format). > > See http://wiki.nginx.org/NginxHttpLogModule for details. > Thank you Maxim. Very much appreciated. I had read that page but I guess I didn't quite "get it". I've always used the standard "combined" format as it had suited my needs until now. It is now logging "HIT" or "MISS" for each request in that location block. It is safe to assume that with fastcgi_cache_min_uses set at 2 that for every first miss shared memory is allocated, that the file is written to the cache on the second miss, and that each logged "hit" is an actual serve from the cache (third and subsequent requests for the file)? > Maxim Dounin > > > Jim From lists at ruby-forum.com Tue Jul 14 08:57:36 2009 From: lists at ruby-forum.com (Tony Payne) Date: Tue, 14 Jul 2009 06:57:36 +0200 Subject: Upload progress module: "track_uploads" directive inside an "if" block. In-Reply-To: <49B0EBCB-7F46-442C-BA0E-A6B4B6CF6D3D@rothlis.net> References: <278DB3CC-5C63-4E68-A1DB-3064EC0279C5@rothlis.net> <4973C85C.8090705@grid.net.ru> <2CBF9BBB-0677-41AD-A051-086EFFA59D79@rothlis.net> <49B0EBCB-7F46-442C-BA0E-A6B4B6CF6D3D@rothlis.net> Message-ID: <87b164d4ff3946aa4c1f3edeb565c570@ruby-forum.com> Dave Rothlisberger wrote: > After further investigation, it seems that the upload progress tracker > is stuck at 'state: starting'. > > (while uploading a large file, if I do a GET for /upload_progress?X- > Progress-ID=xxxx in a separate browser window, I get: > > new Object({ 'state' : 'starting' }) > > even a minute after it has started uploading). > > Once again, this only happens if the "track_uploads" directive is > inside an "if" block; otherwise it works fine. > > Regards, > Dave. Rothlisberger. I think I have found a solution to this issue. I've created a location block that grabs the static file extensions and does the normal processing on them without the track_uploads call. Essentially, this means I can't track uploads to URLs ending in CSS, JS, etc. I can live with that. It also means that full page caching probably won't work, but I don't use that. Here's my new location block: location ~* \.(jpg|gif|png|css|js|htm|html)(\?[0-9]+)?$ { proxy_set_header X-Real-IP $remote_addr; proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for; proxy_set_header Host $http_host; proxy_redirect false; proxy_max_temp_file_size 0; if (!-f $request_filename) { proxy_pass http://mongrel; break; } } -- Posted via http://www.ruby-forum.com/. From is at rambler-co.ru Tue Jul 14 09:21:39 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Tue, 14 Jul 2009 09:21:39 +0400 Subject: Cache questions In-Reply-To: <4A5BF6EE.4080306@gmail.com> References: <4A57D142.2070501@gmail.com> <20090713113321.GH67302@mdounin.ru> <20090713114840.GE9584@rambler-co.ru> <4A5B92BE.60002@gmail.com> <20090713201746.GE15437@rambler-co.ru> <4A5BD50F.1020705@gmail.com> <20090714022959.GR67302@mdounin.ru> <4A5BF6EE.4080306@gmail.com> Message-ID: <20090714052139.GA27340@rambler-co.ru> On Mon, Jul 13, 2009 at 11:09:34PM -0400, Jim Ohlstein wrote: > It is now logging "HIT" or "MISS" for each request in that location block. > > It is safe to assume that with fastcgi_cache_min_uses set at 2 that for > every first miss shared memory is allocated, that the file is written to > the cache on the second miss, and that each logged "hit" is an actual > serve from the cache (third and subsequent requests for the file)? Yes. Currently, there are "MISS", "EXPIRED", "STALE", "UPDATING", and "HIT" states. If you will think out a good name for state when miss is just counted, but not cached, I will make this state. -- Igor Sysoev http://sysoev.ru/en/ From nginx-forum at nginx.us Tue Jul 14 12:07:47 2009 From: nginx-forum at nginx.us (bavery) Date: Tue, 14 Jul 2009 04:07:47 -0400 Subject: problem w/ gzip when hiding Content-Type header from upstream server Message-ID: <69e0c4acac418c641823f567866f9e3c.NginxMailingList@forum.nginx.org> I have the exact situation described in the wiki for the proxy_hide_header command, where my upstream server sends my JS & CSS files with the wrong Content-Type header. The file is currently being sent with a text/html content-type from the upstream server, but I am experiencing some occasional issues with browsers not rendering the CSS and suspect it may possibly be due to not having the correct content-type header, since I have read that this can sometimes cause the problems I am seeing. location /combined.css { proxy_pass http://$proxyforwardto/v/vspfiles/assets/combined.css.asp; proxy_hide_header Content-Type; I have the following line in my config file for specifying which file types gzip should compress. gzip_types text/plain text/css application/x-javascript text/xml application/xml application/xml+rss text/javascript; My problem seems to be that once the Content-Type header is hidden with proxy_hide_header, nginx does not seem to be attempting to determine the correct type itself. My CSS & JS files are simply passed on uncompressed, with no content-type header. Mime-types are configured correctly in my nginx config, but it doesn't seem to make a difference for these particular files. I have tried using proxy_set_header to specifically set the content-type header as text/css, but nginx seems to completely ignore this and continues sending the file through without changing the content-type header. Is there something I am missing here? How can I get nginx to recognize that the .css & .js files it is pulling from the upstream server should still be compressed even though the Content-Type header is being hidden? I am running nginx version 0.8.0. Bryan Posted at Nginx Forum: http://forum.nginx.org/read.php?2,4037,4037#msg-4037 From lists at ruby-forum.com Tue Jul 14 12:32:12 2009 From: lists at ruby-forum.com (Robert Gabriel) Date: Tue, 14 Jul 2009 10:32:12 +0200 Subject: Can't start nginx In-Reply-To: <6ede0e531f82cbd739f53b739d7971a6@ruby-forum.com> References: <808d0ac45d2eeeabbc7dbf3881810c7b@ruby-forum.com> <31f5a14539e362a7ad78e4cf31513f2b@ruby-forum.com> <6ede0e531f82cbd739f53b739d7971a6@ruby-forum.com> Message-ID: <5c84e9d9fa93b84ccab51e3a4e60cf49@ruby-forum.com> Andreas Vieten wrote: > Robert Gabriel wrote: >> Try this: >> >> sh -x /etc/init.d/nginx start >> >> and see what happen or put it in a reply, as you wish. > > > Hello, doesn't work either. I didnt say its gonna work, but you can see where the error is coming from... -- Posted via http://www.ruby-forum.com/. From lists at ruby-forum.com Tue Jul 14 12:33:58 2009 From: lists at ruby-forum.com (Robert Gabriel) Date: Tue, 14 Jul 2009 10:33:58 +0200 Subject: Can't start nginx In-Reply-To: <5c84e9d9fa93b84ccab51e3a4e60cf49@ruby-forum.com> References: <808d0ac45d2eeeabbc7dbf3881810c7b@ruby-forum.com> <31f5a14539e362a7ad78e4cf31513f2b@ruby-forum.com> <6ede0e531f82cbd739f53b739d7971a6@ruby-forum.com> <5c84e9d9fa93b84ccab51e3a4e60cf49@ruby-forum.com> Message-ID: <7844bad4b0a6e140d27ad2fdd62d0e43@ruby-forum.com> Robert Gabriel wrote: > Andreas Vieten wrote: >> Robert Gabriel wrote: >>> Try this: >>> >>> sh -x /etc/init.d/nginx start >>> >>> and see what happen or put it in a reply, as you wish. >> >> >> Hello, doesn't work either. > > I didnt say its gonna work, but you can see where the error is coming > from... Wait, are u using Ubuntu?! If so, then do this first: rm -f /bin/sh; ln -s /bin/bash /bin/sh (as root user) -- Posted via http://www.ruby-forum.com/. From jim.ohlstein at gmail.com Tue Jul 14 17:26:22 2009 From: jim.ohlstein at gmail.com (Jim Ohlstein) Date: Tue, 14 Jul 2009 09:26:22 -0400 Subject: Cache questions In-Reply-To: <20090714052139.GA27340@rambler-co.ru> References: <4A57D142.2070501@gmail.com> <20090713113321.GH67302@mdounin.ru> <20090713114840.GE9584@rambler-co.ru> <4A5B92BE.60002@gmail.com> <20090713201746.GE15437@rambler-co.ru> <4A5BD50F.1020705@gmail.com> <20090714022959.GR67302@mdounin.ru> <4A5BF6EE.4080306@gmail.com> <20090714052139.GA27340@rambler-co.ru> Message-ID: <4A5C877E.30704@gmail.com> Igor Sysoev wrote: > On Mon, Jul 13, 2009 at 11:09:34PM -0400, Jim Ohlstein wrote: > > >> It is now logging "HIT" or "MISS" for each request in that location block. >> >> It is safe to assume that with fastcgi_cache_min_uses set at 2 that for >> every first miss shared memory is allocated, that the file is written to >> the cache on the second miss, and that each logged "hit" is an actual >> serve from the cache (third and subsequent requests for the file)? >> > > Yes. Currently, there are "MISS", "EXPIRED", "STALE", "UPDATING", and "HIT" > states. If you will think out a good name for state when miss is just counted, > but not cached, I will make this state. > > > "MISS-NOT_CACHED" fairly well tells the story, if it's not too long. Otherwise "MISS-NC". Or perhaps "MISS-COUNTING". Jim From is at rambler-co.ru Tue Jul 14 19:11:12 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Tue, 14 Jul 2009 19:11:12 +0400 Subject: Cache questions In-Reply-To: <4A5C877E.30704@gmail.com> References: <4A57D142.2070501@gmail.com> <20090713113321.GH67302@mdounin.ru> <20090713114840.GE9584@rambler-co.ru> <4A5B92BE.60002@gmail.com> <20090713201746.GE15437@rambler-co.ru> <4A5BD50F.1020705@gmail.com> <20090714022959.GR67302@mdounin.ru> <4A5BF6EE.4080306@gmail.com> <20090714052139.GA27340@rambler-co.ru> <4A5C877E.30704@gmail.com> Message-ID: <20090714151112.GH31349@rambler-co.ru> On Tue, Jul 14, 2009 at 09:26:22AM -0400, Jim Ohlstein wrote: > > > Igor Sysoev wrote: > >On Mon, Jul 13, 2009 at 11:09:34PM -0400, Jim Ohlstein wrote: > > > > > >>It is now logging "HIT" or "MISS" for each request in that location block. > >> > >>It is safe to assume that with fastcgi_cache_min_uses set at 2 that for > >>every first miss shared memory is allocated, that the file is written to > >>the cache on the second miss, and that each logged "hit" is an actual > >>serve from the cache (third and subsequent requests for the file)? > >> > > > >Yes. Currently, there are "MISS", "EXPIRED", "STALE", "UPDATING", and "HIT" > >states. If you will think out a good name for state when miss is just > >counted, > >but not cached, I will make this state. > > > > > > > "MISS-NOT_CACHED" fairly well tells the story, if it's not too long. > Otherwise "MISS-NC". > > Or perhaps "MISS-COUNTING". May be just "COUNTING" ? -- Igor Sysoev http://sysoev.ru/en/ From jim.ohlstein at gmail.com Tue Jul 14 19:28:15 2009 From: jim.ohlstein at gmail.com (Jim Ohlstein) Date: Tue, 14 Jul 2009 11:28:15 -0400 Subject: Cache questions In-Reply-To: <20090714151112.GH31349@rambler-co.ru> References: <4A57D142.2070501@gmail.com> <20090713113321.GH67302@mdounin.ru> <20090713114840.GE9584@rambler-co.ru> <4A5B92BE.60002@gmail.com> <20090713201746.GE15437@rambler-co.ru> <4A5BD50F.1020705@gmail.com> <20090714022959.GR67302@mdounin.ru> <4A5BF6EE.4080306@gmail.com> <20090714052139.GA27340@rambler-co.ru> <4A5C877E.30704@gmail.com> <20090714151112.GH31349@rambler-co.ru> Message-ID: <4A5CA40F.5030809@gmail.com> Igor Sysoev wrote: > On Tue, Jul 14, 2009 at 09:26:22AM -0400, Jim Ohlstein wrote: > > >> Igor Sysoev wrote: >> >>> On Mon, Jul 13, 2009 at 11:09:34PM -0400, Jim Ohlstein wrote: >>> >>> >>> >>>> It is now logging "HIT" or "MISS" for each request in that location block. >>>> >>>> It is safe to assume that with fastcgi_cache_min_uses set at 2 that for >>>> every first miss shared memory is allocated, that the file is written to >>>> the cache on the second miss, and that each logged "hit" is an actual >>>> serve from the cache (third and subsequent requests for the file)? >>>> >>>> >>> Yes. Currently, there are "MISS", "EXPIRED", "STALE", "UPDATING", and "HIT" >>> states. If you will think out a good name for state when miss is just >>> counted, >>> but not cached, I will make this state. >>> >>> >>> >>> >> "MISS-NOT_CACHED" fairly well tells the story, if it's not too long. >> Otherwise "MISS-NC". >> >> Or perhaps "MISS-COUNTING". >> > > May be just "COUNTING" ? > > So the first miss would be "MISS", subsequent misses would be "COUNTING"? Or vice versa? Either way would be good for me. Jim From is at rambler-co.ru Tue Jul 14 19:41:42 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Tue, 14 Jul 2009 19:41:42 +0400 Subject: Cache questions In-Reply-To: <4A5CA40F.5030809@gmail.com> References: <20090713114840.GE9584@rambler-co.ru> <4A5B92BE.60002@gmail.com> <20090713201746.GE15437@rambler-co.ru> <4A5BD50F.1020705@gmail.com> <20090714022959.GR67302@mdounin.ru> <4A5BF6EE.4080306@gmail.com> <20090714052139.GA27340@rambler-co.ru> <4A5C877E.30704@gmail.com> <20090714151112.GH31349@rambler-co.ru> <4A5CA40F.5030809@gmail.com> Message-ID: <20090714154142.GI31349@rambler-co.ru> On Tue, Jul 14, 2009 at 11:28:15AM -0400, Jim Ohlstein wrote: > > > Igor Sysoev wrote: > >On Tue, Jul 14, 2009 at 09:26:22AM -0400, Jim Ohlstein wrote: > > > > > >>Igor Sysoev wrote: > >> > >>>On Mon, Jul 13, 2009 at 11:09:34PM -0400, Jim Ohlstein wrote: > >>> > >>> > >>> > >>>>It is now logging "HIT" or "MISS" for each request in that location > >>>>block. > >>>> > >>>>It is safe to assume that with fastcgi_cache_min_uses set at 2 that for > >>>>every first miss shared memory is allocated, that the file is written > >>>>to the cache on the second miss, and that each logged "hit" is an > >>>>actual serve from the cache (third and subsequent requests for the > >>>>file)? > >>>> > >>>> > >>>Yes. Currently, there are "MISS", "EXPIRED", "STALE", "UPDATING", and > >>>"HIT" > >>>states. If you will think out a good name for state when miss is just > >>>counted, > >>>but not cached, I will make this state. > >>> > >>> > >>> > >>> > >>"MISS-NOT_CACHED" fairly well tells the story, if it's not too long. > >>Otherwise "MISS-NC". > >> > >>Or perhaps "MISS-COUNTING". > >> > > > >May be just "COUNTING" ? > > > > > So the first miss would be "MISS", subsequent misses would be > "COUNTING"? Or vice versa? Either way would be good for me. No, last miss will be MISS, while the previous ones will be COUNTING. In case "..._cache_min_uses 1" the first will be MISS, since it's the last miss. In "..._cache_min_uses 2": COUNTING, MISS. -- Igor Sysoev http://sysoev.ru/en/ From jim.ohlstein at gmail.com Tue Jul 14 19:55:46 2009 From: jim.ohlstein at gmail.com (Jim Ohlstein) Date: Tue, 14 Jul 2009 11:55:46 -0400 Subject: Cache questions In-Reply-To: <20090714154142.GI31349@rambler-co.ru> References: <20090713114840.GE9584@rambler-co.ru> <4A5B92BE.60002@gmail.com> <20090713201746.GE15437@rambler-co.ru> <4A5BD50F.1020705@gmail.com> <20090714022959.GR67302@mdounin.ru> <4A5BF6EE.4080306@gmail.com> <20090714052139.GA27340@rambler-co.ru> <4A5C877E.30704@gmail.com> <20090714151112.GH31349@rambler-co.ru> <4A5CA40F.5030809@gmail.com> <20090714154142.GI31349@rambler-co.ru> Message-ID: <4A5CAA82.9030901@gmail.com> Igor Sysoev wrote: > No, last miss will be MISS, while the previous ones will be COUNTING. > In case "..._cache_min_uses 1" the first will be MISS, since it's > the last miss. In "..._cache_min_uses 2": COUNTING, MISS. > > Sounds good. Thanks Igor. Jim From nginx-forum at nginx.us Tue Jul 14 20:27:57 2009 From: nginx-forum at nginx.us (mhi) Date: Tue, 14 Jul 2009 12:27:57 -0400 Subject: php sites crash in windows Message-ID: <85e0bbfa73ccb6ae58eef726807a7651.NginxMailingList@forum.nginx.org> few clicks and php sites crash. ab benchmarks get this response every time. Benchmarking 192.168.1.103 (be patient) apr_poll: The timeout specified has expired (70007) Total of 10 requests completed 200 requests or more goes to nginx-php failure. Am I do something wrong or this is a bug? I tested nginx 0.761 and 0.84 with php 5.3 and 5.2.10 on win XP and 2003 php error: Warning: strftime() : It is not safe to rely on the system's timezone settings. You are *required* to use the date.timezone setting or the date_default_timezone_set() function. In case you used any of those methods and you are still getting this warning, you most likely misspelled the timezone identifier. We selected 'Europe/Helsinki' for '3.0/DST' instead in C:\mynginxp\html\pmwiki\pmwiki.php on line 322 First time after restart server it seems ok. nginx.conf #user nobody; worker_processes 2; error_log logs/error.log; events { worker_connections 1024; } http { server_names_hash_bucket_size 64; include mime.types; default_type application/octet-stream; # include fastcgi_params; autoindex on; autoindex_exact_size on; autoindex_localtime off; sendfile on; #tcp_nopush on; keepalive_timeout 0; ignore_invalid_headers on; server { listen 8080; # Replace this IP and port with the right ones for your requirements server_name localhost; # Multiple hostnames seperated by spaces. Replace these as well. location / { root html; index index.php index.html index.htm; } #error_page 404 /404.html; # redirect server error pages to the static page /50x.html # error_page 500 502 503 504 /50x.html; location = /50x.html { root html; } # proxy the PHP scripts to Apache listening on 127.0.0.1:80 # #location ~ \.php$ { # proxy_pass http://127.0.0.1; #} # pass the PHP scripts to FastCGI server listening on 127.0.0.1:9000 # location ~ \.php$ { root html; fastcgi_pass 127.0.0.1:9000; fastcgi_index index.php; fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name; include fastcgi_params; allow 127.0.0.1; } } # another virtual host using mix of IP-, name-, and port-based configuration # server { listen 8080; # listen somename:8080; server_name pmwiki alias localhost.alias; # access_log logs/pmwiki.access.log main; location / { root html; index index.php index.html index.htm; } location ~ \.php$ { root html/pmwiki; fastcgi_pass 127.0.0.1:9000; fastcgi_index index.php; fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name; include fastcgi_params; allow 192.168.1.0/24; } } } Posted at Nginx Forum: http://forum.nginx.org/read.php?2,4046,4046#msg-4046 From iberkner at gmail.com Tue Jul 14 22:14:11 2009 From: iberkner at gmail.com (Ilan Berkner) Date: Tue, 14 Jul 2009 14:14:11 -0400 Subject: setting up ssl ... can the same server section listen on both ports 80 and 443 or do I need to setup a separate server section? Message-ID: <77432c470907141114w7fbe080fj3e3608216138c813@mail.gmail.com> -------------- next part -------------- An HTML attachment was scrubbed... URL: From azifali at gmail.com Tue Jul 14 22:37:45 2009 From: azifali at gmail.com (Asif Ali) Date: Wed, 15 Jul 2009 00:07:45 +0530 Subject: parsing special characters Message-ID: <52d8feb90907141137o524c8851k96302f6c6dc25673@mail.gmail.com> Dear All, How does one escape URL based special characters in nginx? To give you an example a domain can get requests which passes the urls from another domain... For e.g. http://mydomain.com/u=http://anotherdomain.com/someurl/%7c the u paramter can contain special chacters from another url Suggestions are welcome. Thanks in advance regards Asif Ali -------------- next part -------------- An HTML attachment was scrubbed... URL: From maccaday at gmail.com Tue Jul 14 23:23:13 2009 From: maccaday at gmail.com (Marcus Clyne) Date: Tue, 14 Jul 2009 22:23:13 +0300 Subject: Cache questions In-Reply-To: <4A5C877E.30704@gmail.com> References: <4A57D142.2070501@gmail.com> <20090713113321.GH67302@mdounin.ru> <20090713114840.GE9584@rambler-co.ru> <4A5B92BE.60002@gmail.com> <20090713201746.GE15437@rambler-co.ru> <4A5BD50F.1020705@gmail.com> <20090714022959.GR67302@mdounin.ru> <4A5BF6EE.4080306@gmail.com> <20090714052139.GA27340@rambler-co.ru> <4A5C877E.30704@gmail.com> Message-ID: <4A5CDB21.5020809@gmail.com> Jim Ohlstein wrote: > > > Igor Sysoev wrote: >> On Mon, Jul 13, 2009 at 11:09:34PM -0400, Jim Ohlstein wrote: >> >> >>> It is now logging "HIT" or "MISS" for each request in that location >>> block. >>> >>> It is safe to assume that with fastcgi_cache_min_uses set at 2 that >>> for every first miss shared memory is allocated, that the file is >>> written to the cache on the second miss, and that each logged "hit" >>> is an actual serve from the cache (third and subsequent requests for >>> the file)? >>> >> >> Yes. Currently, there are "MISS", "EXPIRED", "STALE", "UPDATING", and >> "HIT" >> states. If you will think out a good name for state when miss is just >> counted, >> but not cached, I will make this state. >> >> >> > "MISS-NOT_CACHED" fairly well tells the story, if it's not too long. > Otherwise "MISS-NC". > > Or perhaps "MISS-COUNTING". > > Jim > > How about "UNCACHED"? Marcus From mike503 at gmail.com Tue Jul 14 23:58:10 2009 From: mike503 at gmail.com (Michael Shadle) Date: Tue, 14 Jul 2009 12:58:10 -0700 Subject: setting up ssl ... can the same server section listen on both ports 80 and 443 or do I need to setup a separate server section? In-Reply-To: <77432c470907141114w7fbe080fj3e3608216138c813@mail.gmail.com> References: <77432c470907141114w7fbe080fj3e3608216138c813@mail.gmail.com> Message-ID: someone just posted a config example in the past couple days about this On Tue, Jul 14, 2009 at 11:14 AM, Ilan Berkner wrote: > > From iberkner at gmail.com Wed Jul 15 00:06:57 2009 From: iberkner at gmail.com (Ilan Berkner) Date: Tue, 14 Jul 2009 16:06:57 -0400 Subject: setting up ssl ... can the same server section listen on both ports 80 and 443 or do I need to setup a separate server section? In-Reply-To: References: <77432c470907141114w7fbe080fj3e3608216138c813@mail.gmail.com> Message-ID: <77432c470907141306s18aafc01i3cc320e6b251317c@mail.gmail.com> If you can point it out to me... I searched, can't find it. On Tue, Jul 14, 2009 at 3:58 PM, Michael Shadle wrote: > someone just posted a config example in the past couple days about this > > On Tue, Jul 14, 2009 at 11:14 AM, Ilan Berkner wrote: > > > > > > -------------- next part -------------- An HTML attachment was scrubbed... URL: From mdounin at mdounin.ru Wed Jul 15 00:16:44 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Wed, 15 Jul 2009 00:16:44 +0400 Subject: setting up ssl ... can the same server section listen on both ports 80 and 443 or do I need to setup a separate server section? In-Reply-To: <77432c470907141306s18aafc01i3cc320e6b251317c@mail.gmail.com> References: <77432c470907141114w7fbe080fj3e3608216138c813@mail.gmail.com> <77432c470907141306s18aafc01i3cc320e6b251317c@mail.gmail.com> Message-ID: <20090714201643.GY67302@mdounin.ru> Hello! On Tue, Jul 14, 2009 at 04:06:57PM -0400, Ilan Berkner wrote: > If you can point it out to me... I searched, can't find it. http://wiki.nginx.org/NginxHttpCoreModule#listen server { listen 80; listen 443 default ssl; ... } Maxim Dounin > > On Tue, Jul 14, 2009 at 3:58 PM, Michael Shadle wrote: > > > someone just posted a config example in the past couple days about this > > > > On Tue, Jul 14, 2009 at 11:14 AM, Ilan Berkner wrote: > > > > > > > > > > From mike503 at gmail.com Wed Jul 15 00:19:43 2009 From: mike503 at gmail.com (Michael Shadle) Date: Tue, 14 Jul 2009 13:19:43 -0700 Subject: setting up ssl ... can the same server section listen on both ports 80 and 443 or do I need to setup a separate server section? In-Reply-To: <77432c470907141306s18aafc01i3cc320e6b251317c@mail.gmail.com> References: <77432c470907141114w7fbe080fj3e3608216138c813@mail.gmail.com> <77432c470907141306s18aafc01i3cc320e6b251317c@mail.gmail.com> Message-ID: http://marc.info/?l=nginx&m=124747329127204&w=2 http://marc.info/?l=nginx&m=124747239025852&w=2 http://marc.info/?l=nginx&m=124747216625615&w=2 On Tue, Jul 14, 2009 at 1:06 PM, Ilan Berkner wrote: > If you can point it out to me... I searched, can't find it. > > On Tue, Jul 14, 2009 at 3:58 PM, Michael Shadle wrote: >> >> someone just posted a config example in the past couple days about this >> >> On Tue, Jul 14, 2009 at 11:14 AM, Ilan Berkner wrote: >> > >> > >> > > From iberkner at gmail.com Wed Jul 15 00:24:24 2009 From: iberkner at gmail.com (Ilan Berkner) Date: Tue, 14 Jul 2009 16:24:24 -0400 Subject: setting up ssl ... can the same server section listen on both ports 80 and 443 or do I need to setup a separate server section? In-Reply-To: References: <77432c470907141114w7fbe080fj3e3608216138c813@mail.gmail.com> <77432c470907141306s18aafc01i3cc320e6b251317c@mail.gmail.com> Message-ID: <77432c470907141324y568d1069mcd0e6c1cb6350e18@mail.gmail.com> Thanks Maxim and Michael, I got it. On Tue, Jul 14, 2009 at 4:19 PM, Michael Shadle wrote: > http://marc.info/?l=nginx&m=124747329127204&w=2 > http://marc.info/?l=nginx&m=124747239025852&w=2 > http://marc.info/?l=nginx&m=124747216625615&w=2 > > On Tue, Jul 14, 2009 at 1:06 PM, Ilan Berkner wrote: > > If you can point it out to me... I searched, can't find it. > > > > On Tue, Jul 14, 2009 at 3:58 PM, Michael Shadle > wrote: > >> > >> someone just posted a config example in the past couple days about this > >> > >> On Tue, Jul 14, 2009 at 11:14 AM, Ilan Berkner > wrote: > >> > > >> > > >> > > > > > > -------------- next part -------------- An HTML attachment was scrubbed... URL: From mike503 at gmail.com Wed Jul 15 00:25:00 2009 From: mike503 at gmail.com (Michael Shadle) Date: Tue, 14 Jul 2009 13:25:00 -0700 Subject: setting up ssl ... can the same server section listen on both ports 80 and 443 or do I need to setup a separate server section? In-Reply-To: <20090714201643.GY67302@mdounin.ru> References: <77432c470907141114w7fbe080fj3e3608216138c813@mail.gmail.com> <77432c470907141306s18aafc01i3cc320e6b251317c@mail.gmail.com> <20090714201643.GY67302@mdounin.ru> Message-ID: ideally it would be listen 1.2.3.4:443 default ssl; right? otherwise this block will be the default ssl host for anything else unmatched. i typically use ip:443 in all my ssl stuff to ensure i am only going to serve up the cert on the proper sites anyway. On Tue, Jul 14, 2009 at 1:16 PM, Maxim Dounin wrote: > Hello! > > On Tue, Jul 14, 2009 at 04:06:57PM -0400, Ilan Berkner wrote: > >> If you can point it out to me... I searched, can't find it. > > http://wiki.nginx.org/NginxHttpCoreModule#listen > > server { > ? ?listen 80; > ? ?listen 443 default ssl; > > ? ?... > } > > Maxim Dounin > > >> >> On Tue, Jul 14, 2009 at 3:58 PM, Michael Shadle wrote: >> >> > someone just posted a config example in the past couple days about this >> > >> > On Tue, Jul 14, 2009 at 11:14 AM, Ilan Berkner wrote: >> > > >> > > >> > >> > > > From iberkner at gmail.com Wed Jul 15 00:33:01 2009 From: iberkner at gmail.com (Ilan Berkner) Date: Tue, 14 Jul 2009 16:33:01 -0400 Subject: setting up ssl ... can the same server section listen on both ports 80 and 443 or do I need to setup a separate server section? In-Reply-To: References: <77432c470907141114w7fbe080fj3e3608216138c813@mail.gmail.com> <77432c470907141306s18aafc01i3cc320e6b251317c@mail.gmail.com> <20090714201643.GY67302@mdounin.ru> Message-ID: <77432c470907141333j7ff79f78y3eb0286184fc2df8@mail.gmail.com> That's the way I'll set it up. I specify IP as well. On Tue, Jul 14, 2009 at 4:25 PM, Michael Shadle wrote: > ideally it would be listen 1.2.3.4:443 default ssl; right? > > otherwise this block will be the default ssl host for anything else > unmatched. > > i typically use ip:443 in all my ssl stuff to ensure i am only going > to serve up the cert on the proper sites anyway. > > On Tue, Jul 14, 2009 at 1:16 PM, Maxim Dounin wrote: > > Hello! > > > > On Tue, Jul 14, 2009 at 04:06:57PM -0400, Ilan Berkner wrote: > > > >> If you can point it out to me... I searched, can't find it. > > > > http://wiki.nginx.org/NginxHttpCoreModule#listen > > > > server { > > listen 80; > > listen 443 default ssl; > > > > ... > > } > > > > Maxim Dounin > > > > > >> > >> On Tue, Jul 14, 2009 at 3:58 PM, Michael Shadle > wrote: > >> > >> > someone just posted a config example in the past couple days about > this > >> > > >> > On Tue, Jul 14, 2009 at 11:14 AM, Ilan Berkner > wrote: > >> > > > >> > > > >> > > >> > > > > > > > -------------- next part -------------- An HTML attachment was scrubbed... URL: From maccaday at gmail.com Wed Jul 15 00:56:27 2009 From: maccaday at gmail.com (Marcus Clyne) Date: Tue, 14 Jul 2009 23:56:27 +0300 Subject: Why u_char* not char* In-Reply-To: <77432c470907141306s18aafc01i3cc320e6b251317c@mail.gmail.com> References: <77432c470907141114w7fbe080fj3e3608216138c813@mail.gmail.com> <77432c470907141306s18aafc01i3cc320e6b251317c@mail.gmail.com> Message-ID: <4A5CF0FB.1040409@gmail.com> Hi, Why are strings in Nginx stored as u_char*'s and not char*'s pointers? What's the advantage? Thanks, Marcus. From lists at ruby-forum.com Wed Jul 15 01:17:31 2009 From: lists at ruby-forum.com (Luka Horvatic) Date: Tue, 14 Jul 2009 23:17:31 +0200 Subject: A few questions Message-ID: <8c3b46bcf0e53298829c2153767cd5d6@ruby-forum.com> First,for some reason virtual host which i create,i always get error 403.I tried many things,from chowing to various owners and other,until i found temporal solution-i created copy of html directory(which is created during install and which working fine) called bla,and pointed domain to bla directory and then it worked.But this is problematic since all files are in home directory,which means i will need to copy all of them into /usr/local/nginx/. Also what is problem is with alias,i have a script which is located at /home/ also and that again causing that 403 error since it's not located in /usr/local/nginx.I tried to make it work by creating symbolic link but that didn't worked.Seems only solution would be to copy script to nginx directory which will require a lot of file editing because of location change. Also there is a problem with domain alias-*.domain.com not working at all,nor 1.domain.com.If i have *.domain.com instead domain.com,then i get 404 error on directory alias,while it works on www. How do i enable hotlink blank protection for images,where images can be loaded only by specified domains and direct access? And how to rewrite following apache rule: RewriteEngine on RewriteRule ^directory/(.*)$ /st/st.php?%{QUERY_STRING} I using latest stable version,and i think same problems was on latest legacy version too. Here is my server section of domain: server { listen 81; server_name www.domain.com *.domain.com ; #charset koi8-r; #access_log logs/host.access.log main; location / { root bla; index index.php index.html index.htm; } location /TC/ { alias /home/tradecontrol/client/; } #error_page 404 /404.html; # redirect server error pages to the static page /50x.html # error_page 500 502 503 504 /50x.html; location = /50x.html { root bla; } # proxy the PHP scripts to Apache listening on 127.0.0.1:80 # #location ~ \.php$ { # proxy_pass http://127.0.0.1; #} # pass the PHP scripts to FastCGI server listening on 127.0.0.1:9000 # location ~ \.php$ { root bla; fastcgi_pass 127.0.0.1:9000; fastcgi_index index.php; fastcgi_param SCRIPT_FILENAME /usr/local/nginx/bla$fastcgi_script_name; include fastcgi_params; } # deny access to .htaccess files, if Apache's document root # concurs with nginx's one # #location ~ /\.ht { # deny all; #} } -- Posted via http://www.ruby-forum.com/. From merlincorey at dc949.org Wed Jul 15 01:33:57 2009 From: merlincorey at dc949.org (merlin corey) Date: Tue, 14 Jul 2009 14:33:57 -0700 Subject: parsing special characters In-Reply-To: <52d8feb90907141137o524c8851k96302f6c6dc25673@mail.gmail.com> References: <52d8feb90907141137o524c8851k96302f6c6dc25673@mail.gmail.com> Message-ID: Asif, I am sorry, I am confused. Would you like to encode the string/URI for example to use in a rewrite directive, or do you want to decode it for use in a regular expression capture? Thanks, Merlin On Tue, Jul 14, 2009 at 11:37 AM, Asif Ali wrote: > Dear All, > How does one escape URL based special characters in nginx? > To give you an example > a domain can get requests which passes the urls from another domain... > For e.g. > http://mydomain.com/u=http://anotherdomain.com/someurl/%7c > the u paramter can contain special chacters from another url > Suggestions are welcome. Thanks in advance > regards > Asif Ali > From merlincorey at dc949.org Wed Jul 15 01:37:58 2009 From: merlincorey at dc949.org (merlin corey) Date: Tue, 14 Jul 2009 14:37:58 -0700 Subject: Cache questions In-Reply-To: <4A5CDB21.5020809@gmail.com> References: <4A57D142.2070501@gmail.com> <20090713114840.GE9584@rambler-co.ru> <4A5B92BE.60002@gmail.com> <20090713201746.GE15437@rambler-co.ru> <4A5BD50F.1020705@gmail.com> <20090714022959.GR67302@mdounin.ru> <4A5BF6EE.4080306@gmail.com> <20090714052139.GA27340@rambler-co.ru> <4A5C877E.30704@gmail.com> <4A5CDB21.5020809@gmail.com> Message-ID: COUNTING made good sense to me and was suitably succinct, and I have no proble with MISSED-NOTCACHED or MISSED-NC. UNCACHED seems good too, but it could be confusing some new (and old) that this means it was removed from cache, I think. 2009/7/14 Marcus Clyne : > Jim Ohlstein wrote: >> >> >> Igor Sysoev wrote: >>> >>> On Mon, Jul 13, 2009 at 11:09:34PM -0400, Jim Ohlstein wrote: >>> >>> >>>> >>>> It is now logging "HIT" or "MISS" for each request in that location >>>> block. >>>> >>>> It is safe to assume that with fastcgi_cache_min_uses set at 2 that for >>>> every first miss shared memory is allocated, that the file is written to the >>>> cache on the second miss, and that each logged "hit" is an actual serve from >>>> the cache (third and subsequent requests for the file)? >>>> >>> >>> Yes. Currently, there are "MISS", "EXPIRED", "STALE", "UPDATING", and >>> "HIT" >>> states. If you will think out a good name for state when miss is just >>> counted, >>> but not cached, I will make this state. >>> >>> >>> >> >> "MISS-NOT_CACHED" ?fairly well tells the story, if it's not too long. >> Otherwise ?"MISS-NC". >> >> Or perhaps "MISS-COUNTING". >> >> Jim >> >> > How about "UNCACHED"? > > Marcus > > From merlincorey at dc949.org Wed Jul 15 01:40:50 2009 From: merlincorey at dc949.org (merlin corey) Date: Tue, 14 Jul 2009 14:40:50 -0700 Subject: Why u_char* not char* In-Reply-To: <4A5CF0FB.1040409@gmail.com> References: <77432c470907141114w7fbe080fj3e3608216138c813@mail.gmail.com> <77432c470907141306s18aafc01i3cc320e6b251317c@mail.gmail.com> <4A5CF0FB.1040409@gmail.com> Message-ID: Hi, What's the difference in storage of these types in memory? Thanks, Merlin On Tue, Jul 14, 2009 at 1:56 PM, Marcus Clyne wrote: > Hi, > > Why are strings in Nginx stored as u_char*'s and not char*'s pointers? > ?What's the advantage? > > Thanks, > > Marcus. > > From nginx at tomvalentine.net Wed Jul 15 01:47:04 2009 From: nginx at tomvalentine.net (nginx at tomvalentine.net) Date: Tue, 14 Jul 2009 22:47:04 +0100 Subject: auth_basic password problems Message-ID: <4A5CFCD8.1040204@tomvalentine.net> Hi The authentication on windows now seems to work, to an extent. No more 500 errors! --> thanks! However, I can not seem to get it to recognise the password supplied by the user unless the user supplies the hashed password. The attached htpasswd file was created by htpasswd in Linux (htpasswd in Windows doesn't do crypt); and nginx 0.6.32 running in Linux has no problems with it. But nginx 0.8.5 in Vista wont accept 'password' as the password, it will only accept 'xGtGE5iXWkYCI' E.g. htpasswd -db /var/www/.htpasswd nginx2 password produced nginx2:xGtGE5iXWkYCI and nginx accepts not 'password' but 'xGtGE5iXWkYCI' in Windows. Sorry, couldn't be bothered compiling nginx for Linux (I need sleep now) to see if it was Windows specific or version specific. Any ideas? Tom -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: .htpasswd.txt URL: From maccaday at gmail.com Wed Jul 15 02:02:48 2009 From: maccaday at gmail.com (Marcus Clyne) Date: Wed, 15 Jul 2009 01:02:48 +0300 Subject: Why u_char* not char* In-Reply-To: References: <77432c470907141114w7fbe080fj3e3608216138c813@mail.gmail.com> <77432c470907141306s18aafc01i3cc320e6b251317c@mail.gmail.com> <4A5CF0FB.1040409@gmail.com> Message-ID: <4A5D0088.4000101@gmail.com> Hi, merlin corey wrote: > What's the difference in storage of these types in memory? > AFAIK, none - u_char is just an unsigned char. I can't see why any system would store them differently and I'm guessing that type conversion would be a noop on every system (though I'd be interested to hear of any evidence otherwise). How they are treated when representing them as numbers or when they're involved with any kind of mathematical operations would be different, though (in general). The pain in using u_char* for strings vs char* is that you often have to deal with typecasting when interacting with other functions, reading/writing etc. It's much rarer that other functions (especially system ones) deal with u_char's rather than char's. I was wondering why bother with the extra hassle. I'm guessing there's a reason why Igor chose them over char*'s, but I don't know what it is. Cheers, Marcus. From nunomagalhaes at eu.ipp.pt Wed Jul 15 02:07:02 2009 From: nunomagalhaes at eu.ipp.pt (=?UTF-8?Q?Nuno_Magalh=C3=A3es?=) Date: Tue, 14 Jul 2009 23:07:02 +0100 Subject: A few questions In-Reply-To: <8c3b46bcf0e53298829c2153767cd5d6@ruby-forum.com> References: <8c3b46bcf0e53298829c2153767cd5d6@ruby-forum.com> Message-ID: <6b1504c40907141507q2bcee998re6c3cc237b0e3e1c@mail.gmail.com> On Tue, Jul 14, 2009 at 22:17, Luka Horvatic wrote: > First,for some reason virtual host which i create,i always get error > 403. [...] About your first paragraph: The HTTP 403 Forbidden, as you've already undoubtedly searched, means the server won't let the client access that resource. One common issue is permissions indeed: did you try to 777 everything, including the pertaining directory? Another problem is accessing directories (domain.com/foo/) that do not have an index.* file. If the server is configured not to list the contents of that directory, it'll send 403. server { listen 81; server_name www.domain.com *.domain.com ; You don't need the *. location / { root bla; Should be root /bla, starting from your system's / RTFM the rest -- () ascii ribbon campaign - against html e-mail /\ ascii-rubanda kampajno - kontra? html-a retpo?to From rhodes.daniel at gmail.com Wed Jul 15 02:41:58 2009 From: rhodes.daniel at gmail.com (Daniel Rhodes) Date: Tue, 14 Jul 2009 15:41:58 -0700 Subject: Problem with flv module in Ubuntu Message-ID: <2f2bd9a50907141541p21d3e63dl97b273b1e945c24e@mail.gmail.com> Hello, I have installed Nginx from source (tried both version 0.7.47 and 0.7.61) on an Ubuntu Hardy 64-bit machine, and in both cases, the flv module does not work. I am using the JW Player to stream the files and it only sends one query string parameter, which is what is supposed to work with Nginx's FLV module. When I try seeking, I get this response: 2009/07/14 18:29:40 [info] 31340#0: *644 client closed prematurely connection while sending response to client, client: 98.210.13.200, server: xxxxxx, request: "GET /content/b/6/88860.flv?start=10482045 HTTP/1.1", host: "xxxxxx" The config file is: location ~* ^.+.(flv)$ { root /opt/www/html; gzip off; expires 30d; limit_rate 500k; flv; } The configure information is: nginx version: nginx/0.7.47 built by gcc 4.2.4 (Ubuntu 4.2.4-1ubuntu3) configure arguments: --with-http_flv_module --prefix=/usr/local --with-http_gzip_static_module --with-poll_module --with-http_secure_link_module --with-http_stub_status_module Any help would be much appreciated! Thanks, Dan From mohammad_ysm at yahoo.com Wed Jul 15 04:01:31 2009 From: mohammad_ysm at yahoo.com (Mohammad Kolahdouzan) Date: Tue, 14 Jul 2009 17:01:31 -0700 (PDT) Subject: configuration for proxy_pass/upstream combination In-Reply-To: <16b7ceee0906041611r704c24fficc8ff71ba87298a5@mail.gmail.com> References: <16b7ceee0906041611r704c24fficc8ff71ba87298a5@mail.gmail.com> Message-ID: <662723.57598.qm@web110105.mail.gq1.yahoo.com> I have my own module in which when a request is sent to /mymodule, it'll initiate a new subrequest to "/a1" and passes a bunch of new arguments to it. As an example, a call to "http://localhost/mymodule" will generate a subrequest like "http://a.com/?x=1&y=2" I have a config similar to the following to support that : server { location /mymodule { my_module; } location /a1 { proxy_pass http://a.com/; } } Now the issue that I have is when I want to add redundancy. If I change the config file to something similar to the following: upstream server1{ server http://a.com; server http://b.com; } server { location /mymodule { my_module; } location /a1 { proxy_pass http://server1; } } then a request to "http://localhost/mymodule" will generate a subrequest like "http://a.com/a1/?x=1&y=1". I am wondering what it is that I am doing wrong which adds "/a1/" to the subrequest to upstream. Thanks, -M -------------- next part -------------- An HTML attachment was scrubbed... URL: From mdounin at mdounin.ru Wed Jul 15 05:41:14 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Wed, 15 Jul 2009 05:41:14 +0400 Subject: Why u_char* not char* In-Reply-To: <4A5CF0FB.1040409@gmail.com> References: <77432c470907141114w7fbe080fj3e3608216138c813@mail.gmail.com> <77432c470907141306s18aafc01i3cc320e6b251317c@mail.gmail.com> <4A5CF0FB.1040409@gmail.com> Message-ID: <20090715014113.GZ67302@mdounin.ru> Hello! On Tue, Jul 14, 2009 at 11:56:27PM +0300, Marcus Clyne wrote: > Hi, > > Why are strings in Nginx stored as u_char*'s and not char*'s pointers? > What's the advantage? I'm not sure why Igor choose it, but there are at least several reasons to use 'unsigned char' (aka u_char) instead of 'char' (which may be either signed or unsigned): - Constructs like u_char map[] = { 0, 0, 0, 1, 1, ... }; u_char *p; ... if (map[*p]) { ... } work as expected for all possible character values without any extra typecasting. - Comparision works in predictable way. And you will get (mostly) reasonable sorting on any arbitraty data even without collation support. - Overflow behaviour undefined for signed types, and bitwise operators are undefined for negative values. So basically if you deal with abitrary byte streams in some arbitrary way as nginx do - 'unsigned char' is better choice. Maxim Dounin p.s. It's really good idea to start new thread for unrelated questions. From mdounin at mdounin.ru Wed Jul 15 05:55:03 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Wed, 15 Jul 2009 05:55:03 +0400 Subject: configuration for proxy_pass/upstream combination In-Reply-To: <662723.57598.qm@web110105.mail.gq1.yahoo.com> References: <16b7ceee0906041611r704c24fficc8ff71ba87298a5@mail.gmail.com> <662723.57598.qm@web110105.mail.gq1.yahoo.com> Message-ID: <20090715015502.GA67302@mdounin.ru> Hello! On Tue, Jul 14, 2009 at 05:01:31PM -0700, Mohammad Kolahdouzan wrote: > > I have my own module in which when a request is sent to > /mymodule, it'll initiate a new > subrequest to "/a1" and passes a bunch of new arguments to it. As an > example, a call to "http://localhost/mymodule" will generate a > subrequest like "http://a.com/?x=1&y=2" > I have a config similar to the following to support that : > > server { > location /mymodule { > my_module; > } > location /a1 { > proxy_pass http://a.com/; > } > } > > > Now the issue that I have is when I want to add redundancy. If I change the config file to something similar to the following: > > upstream server1{ > server http://a.com; > server http://b.com; > } > server { > location /mymodule { > my_module; > } > location /a1 { > proxy_pass http://server1; - proxy_pass http://server1; + proxy_pass http://server1/; See http://wiki.nginx.org/NginxHttpProxyModule#proxy_pass for details. Maxim Dounin > } > } > > then a request to "http://localhost/mymodule" will generate a subrequest like "http://a.com/a1/?x=1&y=1". I am wondering what it is that I am doing wrong which adds "/a1/" to the subrequest to upstream. > > Thanks, > -M > > > From is at rambler-co.ru Wed Jul 15 11:06:12 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Wed, 15 Jul 2009 11:06:12 +0400 Subject: Why u_char* not char* In-Reply-To: <20090715014113.GZ67302@mdounin.ru> References: <77432c470907141114w7fbe080fj3e3608216138c813@mail.gmail.com> <77432c470907141306s18aafc01i3cc320e6b251317c@mail.gmail.com> <4A5CF0FB.1040409@gmail.com> <20090715014113.GZ67302@mdounin.ru> Message-ID: <20090715070612.GD51384@rambler-co.ru> On Wed, Jul 15, 2009 at 05:41:14AM +0400, Maxim Dounin wrote: > Hello! > > On Tue, Jul 14, 2009 at 11:56:27PM +0300, Marcus Clyne wrote: > > > Hi, > > > > Why are strings in Nginx stored as u_char*'s and not char*'s pointers? > > What's the advantage? > > I'm not sure why Igor choose it, but there are at least several > reasons to use 'unsigned char' (aka u_char) instead of 'char' > (which may be either signed or unsigned): > > - Constructs like > > u_char map[] = { 0, 0, 0, 1, 1, ... }; > u_char *p; > > ... > > if (map[*p]) { ... } > > work as expected for all possible character values without any > extra typecasting. > > - Comparision works in predictable way. And you will get (mostly) > reasonable sorting on any arbitraty data even without collation > support. > > - Overflow behaviour undefined for signed types, and bitwise > operators are undefined for negative values. > > So basically if you deal with abitrary byte streams in some > arbitrary way as nginx do - 'unsigned char' is better choice. Yes. I prefer to think about 'char *' as character stream (which are unsigned by nature), but not as to a stream of small range singed intergers. This way makes evident the comparisions, bitwise operations, etc. However, this way requires typecasts in trivial cases, e.g.: u_char *p = (u_char *) "text"; But these cases are really trivial (as against to comparisons, bitwise operations, etc.), are easy catched by compilers, and these typecasts are just syntax sugar. -- Igor Sysoev http://sysoev.ru/en/ From maccaday at gmail.com Wed Jul 15 14:13:04 2009 From: maccaday at gmail.com (Marcus Clyne) Date: Wed, 15 Jul 2009 13:13:04 +0300 Subject: Why u_char* not char* In-Reply-To: <20090715070612.GD51384@rambler-co.ru> References: <77432c470907141114w7fbe080fj3e3608216138c813@mail.gmail.com> <77432c470907141306s18aafc01i3cc320e6b251317c@mail.gmail.com> <4A5CF0FB.1040409@gmail.com> <20090715014113.GZ67302@mdounin.ru> <20090715070612.GD51384@rambler-co.ru> Message-ID: <4A5DABB0.7020303@gmail.com> Igor Sysoev wrote: > On Wed, Jul 15, 2009 at 05:41:14AM +0400, Maxim Dounin wrote: > > >> Hello! >> >> On Tue, Jul 14, 2009 at 11:56:27PM +0300, Marcus Clyne wrote: >> >> >>> Hi, >>> >>> Why are strings in Nginx stored as u_char*'s and not char*'s pointers? >>> What's the advantage? >>> >> I'm not sure why Igor choose it, but there are at least several >> reasons to use 'unsigned char' (aka u_char) instead of 'char' >> (which may be either signed or unsigned): >> >> - Constructs like >> >> u_char map[] = { 0, 0, 0, 1, 1, ... }; >> u_char *p; >> >> ... >> >> if (map[*p]) { ... } >> >> work as expected for all possible character values without any >> extra typecasting. >> >> - Comparision works in predictable way. And you will get (mostly) >> reasonable sorting on any arbitraty data even without collation >> support. >> >> - Overflow behaviour undefined for signed types, and bitwise >> operators are undefined for negative values. >> >> So basically if you deal with abitrary byte streams in some >> arbitrary way as nginx do - 'unsigned char' is better choice. >> > > Yes. I prefer to think about 'char *' as character stream (which are > unsigned by nature), but not as to a stream of small range singed > intergers. This way makes evident the comparisions, bitwise operations, etc. > However, this way requires typecasts in trivial cases, e.g.: > > u_char *p = (u_char *) "text"; > > But these cases are really trivial (as against to comparisons, > bitwise operations, etc.), are easy catched by compilers, and these > typecasts are just syntax sugar. > > Thanks for the info (both of you). Marcus. From merlincorey at dc949.org Wed Jul 15 15:57:15 2009 From: merlincorey at dc949.org (merlin corey) Date: Wed, 15 Jul 2009 04:57:15 -0700 Subject: Why u_char* not char* In-Reply-To: <4A5DABB0.7020303@gmail.com> References: <77432c470907141114w7fbe080fj3e3608216138c813@mail.gmail.com> <77432c470907141306s18aafc01i3cc320e6b251317c@mail.gmail.com> <4A5CF0FB.1040409@gmail.com> <20090715014113.GZ67302@mdounin.ru> <20090715070612.GD51384@rambler-co.ru> <4A5DABB0.7020303@gmail.com> Message-ID: Indeed, it was illuminating! Thanks, Merlin From miles at tinyapps.org Wed Jul 15 20:25:56 2009 From: miles at tinyapps.org (TinyApps.Org) Date: Wed, 15 Jul 2009 06:25:56 -1000 Subject: Redirect loop / too many redirects Message-ID: <57B78091-3ED0-4DBC-988F-E64A4C01F212@tinyapps.org> I am trying to redirect all requests for /weblog/index.rss to /weblog/index.rss20 with this rule: location /weblog/index.rss { rewrite /weblog/index.rss /weblog/ index.rss20 permanent; } but I end up with an error message in the browser like: Safari: Too many redirects occurred trying to open ?feed://example.org/weblog/index.rss ?. This might occur if you open a page that is redirected to open another page which then is redirected to open the original page. Firefox: Redirect Loop - Firefox has detected that the server is redirecting the request for this address in a way that will never complete. All other redirect rules work - any idea why this one would fail? From gmm at csdoc.com Wed Jul 15 21:02:27 2009 From: gmm at csdoc.com (Gena Makhomed) Date: Wed, 15 Jul 2009 20:02:27 +0300 Subject: Redirect loop / too many redirects In-Reply-To: <57B78091-3ED0-4DBC-988F-E64A4C01F212@tinyapps.org> References: <57B78091-3ED0-4DBC-988F-E64A4C01F212@tinyapps.org> Message-ID: <664722396.20090715200227@csdoc.com> On Wednesday, July 15, 2009 at 19:25:56, TinyApps.Org wrote: TO> location /weblog/index.rss { TO> rewrite /weblog/index.rss /weblog/index.rss20 permanent; TO> } location = /weblog/index.rss { rewrite ^ /weblog/index.rss20 permanent; } TO> Firefox: Redirect Loop - Firefox has detected that the server is TO> redirecting the request for this address in a way that will never TO> complete. TO> All other redirect rules work - any idea why this one would fail? url "/weblog/index.rss20" was matched by location "/weblog/index.rss" location syntax: http://wiki.nginx.org/NginxHttpCoreModule#location -- Best regards, Gena From miles at tinyapps.org Wed Jul 15 21:26:35 2009 From: miles at tinyapps.org (TinyApps.Org) Date: Wed, 15 Jul 2009 07:26:35 -1000 Subject: Redirect loop / too many redirects In-Reply-To: <664722396.20090715200227@csdoc.com> References: <57B78091-3ED0-4DBC-988F-E64A4C01F212@tinyapps.org> <664722396.20090715200227@csdoc.com> Message-ID: <43286079-5C19-4249-8DA8-824434B25AE6@tinyapps.org> Awesome - thanks so much for the answer, Gena! Just adding the equal sign after "location" did the trick! One last question: is the caret after "rewrite" required? I read the location syntax page but still was not clear on what function the caret would serve when it follows "location = wherever { rewrite..." Yours Gratefully, Miles On Jul 15, 2009, at 7:02 AM, Gena Makhomed wrote: > On Wednesday, July 15, 2009 at 19:25:56, TinyApps.Org wrote: > > TO> location /weblog/index.rss { > TO> rewrite /weblog/index.rss /weblog/index.rss20 permanent; > TO> } > > location = /weblog/index.rss { > rewrite ^ /weblog/index.rss20 permanent; > } > > TO> Firefox: Redirect Loop - Firefox has detected that the server is > TO> redirecting the request for this address in a way that will never > TO> complete. > > TO> All other redirect rules work - any idea why this one would fail? > > url "/weblog/index.rss20" was matched by location "/weblog/index.rss" > > location syntax: http://wiki.nginx.org/NginxHttpCoreModule#location > > -- > Best regards, > Gena > > From gmm at csdoc.com Wed Jul 15 22:06:52 2009 From: gmm at csdoc.com (Gena Makhomed) Date: Wed, 15 Jul 2009 21:06:52 +0300 Subject: Redirect loop / too many redirects In-Reply-To: <43286079-5C19-4249-8DA8-824434B25AE6@tinyapps.org> References: <57B78091-3ED0-4DBC-988F-E64A4C01F212@tinyapps.org> <664722396.20090715200227@csdoc.com> <43286079-5C19-4249-8DA8-824434B25AE6@tinyapps.org> Message-ID: <2910337946.20090715210652@csdoc.com> On Wednesday, July 15, 2009 at 20:26:35, TinyApps.Org wrote: TO> Just adding the equal sign after "location" did the trick! TO> One last question: is the caret after "rewrite" required? >> location = /weblog/index.rss { >> rewrite ^ /weblog/index.rss20 permanent; >> } caret is regexp. ( ^ Match the beginning of the line ) http://perldoc.perl.org/perlre.html#Regular-Expressions yes, regexp here required by rewrite syntax: rewrite regex replacement flag http://wiki.nginx.org/NginxHttpRewriteModule#rewrite -- Best regards, Gena From maccaday at gmail.com Wed Jul 15 23:13:26 2009 From: maccaday at gmail.com (Marcus Clyne) Date: Wed, 15 Jul 2009 22:13:26 +0300 Subject: Best way to get $request_filename In-Reply-To: <43286079-5C19-4249-8DA8-824434B25AE6@tinyapps.org> References: <57B78091-3ED0-4DBC-988F-E64A4C01F212@tinyapps.org> <664722396.20090715200227@csdoc.com> <43286079-5C19-4249-8DA8-824434B25AE6@tinyapps.org> Message-ID: <4A5E2A56.4070006@gmail.com> Hi, What's the best way to get the variable $request_filename from a custom module? Currently I save the index of the $request_filename variable and call ngx_http_get_indexed_variable (r, request_filename_var_index), but is there a better way? Thanks, Marcus. From mdounin at mdounin.ru Wed Jul 15 23:37:56 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Wed, 15 Jul 2009 23:37:56 +0400 Subject: Best way to get $request_filename In-Reply-To: <4A5E2A56.4070006@gmail.com> References: <57B78091-3ED0-4DBC-988F-E64A4C01F212@tinyapps.org> <664722396.20090715200227@csdoc.com> <43286079-5C19-4249-8DA8-824434B25AE6@tinyapps.org> <4A5E2A56.4070006@gmail.com> Message-ID: <20090715193756.GC1050@mdounin.ru> Hello! On Wed, Jul 15, 2009 at 10:13:26PM +0300, Marcus Clyne wrote: > Hi, > > What's the best way to get the variable $request_filename from a custom > module? > > Currently I save the index of the $request_filename variable and call > ngx_http_get_indexed_variable (r, request_filename_var_index), but is > there a better way? If you need variable - using indexed one is the best choise. For request filename you may want to consider using ngx_http_map_uri_to_path() directly though. See static module (or gzip static, or flv, or request_filename variable get handler code) for examples. Maxim Dounin From maccaday at gmail.com Thu Jul 16 02:28:13 2009 From: maccaday at gmail.com (Marcus Clyne) Date: Thu, 16 Jul 2009 01:28:13 +0300 Subject: Best way to get $request_filename In-Reply-To: <20090715193756.GC1050@mdounin.ru> References: <57B78091-3ED0-4DBC-988F-E64A4C01F212@tinyapps.org> <664722396.20090715200227@csdoc.com> <43286079-5C19-4249-8DA8-824434B25AE6@tinyapps.org> <4A5E2A56.4070006@gmail.com> <20090715193756.GC1050@mdounin.ru> Message-ID: <4A5E57FD.7080603@gmail.com> Maxim Dounin wrote: > Hello! > > On Wed, Jul 15, 2009 at 10:13:26PM +0300, Marcus Clyne wrote: > > >> Hi, >> >> What's the best way to get the variable $request_filename from a custom >> module? >> >> Currently I save the index of the $request_filename variable and call >> ngx_http_get_indexed_variable (r, request_filename_var_index), but is >> there a better way? >> > > If you need variable - using indexed one is the best choise. > > For request filename you may want to consider using > ngx_http_map_uri_to_path() directly though. See static module (or > gzip static, or flv, or request_filename variable get handler code) > for examples. > > Maxim Dounin > Thanks. From maccaday at gmail.com Thu Jul 16 03:25:28 2009 From: maccaday at gmail.com (Marcus Clyne) Date: Thu, 16 Jul 2009 02:25:28 +0300 Subject: Daemon directory string In-Reply-To: <20090715193756.GC1050@mdounin.ru> References: <57B78091-3ED0-4DBC-988F-E64A4C01F212@tinyapps.org> <664722396.20090715200227@csdoc.com> <43286079-5C19-4249-8DA8-824434B25AE6@tinyapps.org> <4A5E2A56.4070006@gmail.com> <20090715193756.GC1050@mdounin.ru> Message-ID: <4A5E6568.7000307@gmail.com> Hi, What's the best way to get the path of the directory that nginx was called from (as part of module development)? Is this stored in a string somewhere? Thanks, Marcus. From mdounin at mdounin.ru Thu Jul 16 05:21:16 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Thu, 16 Jul 2009 05:21:16 +0400 Subject: Daemon directory string In-Reply-To: <4A5E6568.7000307@gmail.com> References: <57B78091-3ED0-4DBC-988F-E64A4C01F212@tinyapps.org> <664722396.20090715200227@csdoc.com> <43286079-5C19-4249-8DA8-824434B25AE6@tinyapps.org> <4A5E2A56.4070006@gmail.com> <20090715193756.GC1050@mdounin.ru> <4A5E6568.7000307@gmail.com> Message-ID: <20090716012116.GD1050@mdounin.ru> Hello! On Thu, Jul 16, 2009 at 02:25:28AM +0300, Marcus Clyne wrote: > Hi, > > What's the best way to get the path of the directory that nginx was > called from (as part of module development)? Is this stored in a string > somewhere? No. You should use ngx_conf_full_name() to resolve paths to absolute ones. It uses cycle->prefix (or cycle->conf_prefix), and this in turn is set to startup directory if there is no prefix (either compiled in or specified via -p switch). Maxim Dounin From nginx-forum at nginx.us Thu Jul 16 09:31:22 2009 From: nginx-forum at nginx.us (gerryw) Date: Thu, 16 Jul 2009 01:31:22 -0400 Subject: FastCGI application not receiving multi-part data Message-ID: <6474272efa8edf4f503a18f5fe75a2cf.NginxMailingList@forum.nginx.org> Hello All, I'm working on a FastCGI application the needs to support file upload. When a file is uploaded the FastCGI program is not receiving any data or envp other than METHOD. I think I have configured FastCGI correctly in the nginx config file, because other form submissions work. Can anyone give me an idea what I could check? nginx version: latest FastCGI version: latest Thanks, Gerry Posted at Nginx Forum: http://forum.nginx.org/read.php?2,4091,4091#msg-4091 From rhodes.daniel at gmail.com Thu Jul 16 11:17:05 2009 From: rhodes.daniel at gmail.com (Daniel Rhodes) Date: Thu, 16 Jul 2009 00:17:05 -0700 Subject: FastCGI application not receiving multi-part data In-Reply-To: <6474272efa8edf4f503a18f5fe75a2cf.NginxMailingList@forum.nginx.org> References: <6474272efa8edf4f503a18f5fe75a2cf.NginxMailingList@forum.nginx.org> Message-ID: <2f2bd9a50907160017r2dc77dddgecefd4324cb50f77@mail.gmail.com> Is your FastCGI running on a socket or a port? On Linux, it should be a socket (e.g. unix:/tmp/php.sock) , but on FreeBSD it should be a port (e.g. 127.0.0.1:60000). I found this to be the cause of uploads not being passed back through fastcgi one time. -- DR On Wed, Jul 15, 2009 at 10:31 PM, gerryw wrote: > Hello All, > > I'm working on a FastCGI application the needs to support file upload. When a file is uploaded the FastCGI program is not receiving any data or envp other than METHOD. I think I have configured FastCGI correctly in the nginx config file, because other form submissions work. ?Can anyone give me an idea what I could check? > > nginx version: latest > FastCGI version: latest > > > Thanks, > Gerry > > Posted at Nginx Forum: http://forum.nginx.org/read.php?2,4091,4091#msg-4091 > > > From almir at almirkaric.com Thu Jul 16 11:27:43 2009 From: almir at almirkaric.com (Almir Karic) Date: Thu, 16 Jul 2009 00:27:43 -0700 Subject: FastCGI application not receiving multi-part data In-Reply-To: <6474272efa8edf4f503a18f5fe75a2cf.NginxMailingList@forum.nginx.org> References: <6474272efa8edf4f503a18f5fe75a2cf.NginxMailingList@forum.nginx.org> Message-ID: <2b6a8c330907160027v42a5a519h60583d94f4df1426@mail.gmail.com> I am running django via FCGI that has the file upload working just fine :-), i doubt nginx is source of your problems. Did you remember to add enctype="multipart/form-data"? python/django hacker & sys admin http://almirkaric.com & http://twitter.com/redduck666 On Wed, Jul 15, 2009 at 10:31 PM, gerryw wrote: > Hello All, > > I'm working on a FastCGI application the needs to support file upload. When a file is uploaded the FastCGI program is not receiving any data or envp other than METHOD. I think I have configured FastCGI correctly in the nginx config file, because other form submissions work. ?Can anyone give me an idea what I could check? > > nginx version: latest > FastCGI version: latest > > > Thanks, > Gerry > > Posted at Nginx Forum: http://forum.nginx.org/read.php?2,4091,4091#msg-4091 > > > From miles at tinyapps.org Thu Jul 16 11:46:02 2009 From: miles at tinyapps.org (TinyApps.Org) Date: Wed, 15 Jul 2009 21:46:02 -1000 Subject: Redirect loop / too many redirects In-Reply-To: <2910337946.20090715210652@csdoc.com> References: <57B78091-3ED0-4DBC-988F-E64A4C01F212@tinyapps.org> <664722396.20090715200227@csdoc.com> <43286079-5C19-4249-8DA8-824434B25AE6@tinyapps.org> <2910337946.20090715210652@csdoc.com> Message-ID: <47CB2365-B309-4269-ADB1-A402C19E1C8E@tinyapps.org> Thanks, Gena! On Jul 15, 2009, at 8:06 AM, Gena Makhomed wrote: > On Wednesday, July 15, 2009 at 20:26:35, TinyApps.Org wrote: > > TO> Just adding the equal sign after "location" did the trick! > TO> One last question: is the caret after "rewrite" required? > >>> location = /weblog/index.rss { >>> rewrite ^ /weblog/index.rss20 permanent; >>> } > > caret is regexp. ( ^ Match the beginning of the line ) > > http://perldoc.perl.org/perlre.html#Regular-Expressions > > yes, regexp here required by rewrite syntax: > > rewrite regex replacement flag > > http://wiki.nginx.org/NginxHttpRewriteModule#rewrite > > -- > Best regards, > Gena > > From nginx-forum at nginx.us Thu Jul 16 12:55:00 2009 From: nginx-forum at nginx.us (bavery) Date: Thu, 16 Jul 2009 04:55:00 -0400 Subject: problem w/ gzip when hiding Content-Type header from upstream server In-Reply-To: <69e0c4acac418c641823f567866f9e3c.NginxMailingList@forum.nginx.org> References: <69e0c4acac418c641823f567866f9e3c.NginxMailingList@forum.nginx.org> Message-ID: <925f8f7428ee278a6300710bc88646ab.NginxMailingList@forum.nginx.org> I did some more checking, and am now finding that other .js & .css files from the upstream server are not being compressed as well, except that these files simply are served from the upstream server with no content-type header at all. So why does nginx not recognize the file type it is handling based on mime-type configuration when dealing with proxied files? Bryan Posted at Nginx Forum: http://forum.nginx.org/read.php?2,4037,4098#msg-4098 From nginx-forum at nginx.us Thu Jul 16 12:56:00 2009 From: nginx-forum at nginx.us (gerryw) Date: Thu, 16 Jul 2009 04:56:00 -0400 Subject: FastCGI application not receiving multi-part data In-Reply-To: <6474272efa8edf4f503a18f5fe75a2cf.NginxMailingList@forum.nginx.org> References: <6474272efa8edf4f503a18f5fe75a2cf.NginxMailingList@forum.nginx.org> Message-ID: Hello All, Daniel: >Is your FastCGI running on a socket or a port? On Linux, it should be >a socket (e.g. unix:/tmp/php.sock) , but on FreeBSD it should be a >port (e.g. 127.0.0.1:60000). I found this to be the cause of uploads >not being passed back through fastcgi one time. I was using a port. When I switched to a unix:socket I started getting some data, but it's either garbage or incomplete, because the parser segfaults trying to parse it. Could it be some kind of buffer size issue? Almir: > Did you remember to add enctype="multipart/form-data"? Yes. Thanks, Gerry Posted at Nginx Forum: http://forum.nginx.org/read.php?2,4091,4099#msg-4099 From nginx-forum at nginx.us Thu Jul 16 13:16:16 2009 From: nginx-forum at nginx.us (gerryw) Date: Thu, 16 Jul 2009 05:16:16 -0400 Subject: FastCGI application not receiving multi-part data In-Reply-To: <6474272efa8edf4f503a18f5fe75a2cf.NginxMailingList@forum.nginx.org> References: <6474272efa8edf4f503a18f5fe75a2cf.NginxMailingList@forum.nginx.org> Message-ID: Hello All, In order to eliminate the parser as the problem, I switched to cgicc. Unfortunately, I got the same result. I know that cgicc works, because I've used it with other web servers. I guess I'm still thinking the data is incomplete or doesn't match the length passed. Any thoughts? Thanks, Gerry Posted at Nginx Forum: http://forum.nginx.org/read.php?2,4091,4100#msg-4100 From rhodes.daniel at gmail.com Thu Jul 16 13:24:59 2009 From: rhodes.daniel at gmail.com (Daniel Rhodes) Date: Thu, 16 Jul 2009 02:24:59 -0700 Subject: FastCGI application not receiving multi-part data In-Reply-To: References: <6474272efa8edf4f503a18f5fe75a2cf.NginxMailingList@forum.nginx.org> Message-ID: <2f2bd9a50907160224od8c29d4ge5ebc2ab6c83c5c7@mail.gmail.com> I'm not familiar with that particular problem, although here are the settings I use for the fcgi buffers (which has worked with me for files > 1GB being passed on to PHP): fastcgi_connect_timeout 60; fastcgi_send_timeout 180; fastcgi_read_timeout 180; fastcgi_buffer_size 128k; fastcgi_buffers 4 256k; fastcgi_busy_buffers_size 256k; fastcgi_temp_file_write_size 256k; fastcgi_intercept_errors on; fastcgi_pass_header *; There are also other values, like the document body size which should be set. You can find information about that in the documentation. On Thu, Jul 16, 2009 at 1:56 AM, gerryw wrote: > Hello All, > > Daniel: >>Is your FastCGI running on a socket or a port? On Linux, it should be >>a socket (e.g. unix:/tmp/php.sock) , but on FreeBSD it should be a >>port (e.g. 127.0.0.1:60000). I found this to be the cause of uploads >>not being passed back through fastcgi one time. > > I was using a port. When I switched to a unix:socket I started getting some data, but it's either garbage or incomplete, because the parser segfaults trying to parse it. Could it be some kind of buffer size issue? > > Almir: >> Did you remember to add enctype="multipart/form-data"? > Yes. > > Thanks, > Gerry > > Posted at Nginx Forum: http://forum.nginx.org/read.php?2,4091,4099#msg-4099 > > > From nginx-forum at nginx.us Thu Jul 16 13:27:31 2009 From: nginx-forum at nginx.us (gerryw) Date: Thu, 16 Jul 2009 05:27:31 -0400 Subject: FastCGI application not receiving multi-part data In-Reply-To: <6474272efa8edf4f503a18f5fe75a2cf.NginxMailingList@forum.nginx.org> References: <6474272efa8edf4f503a18f5fe75a2cf.NginxMailingList@forum.nginx.org> Message-ID: <14a2ed3e7b6b55d3caacbc706609cefb.NginxMailingList@forum.nginx.org> Hello All, I also did a test with just the vanilla FastCGI API calls. The only env related variable I was able to get was "REQUEST_METHOD". None of the rest were set. Thanks, Gerry Posted at Nginx Forum: http://forum.nginx.org/read.php?2,4091,4101#msg-4101 From support at oeko.net Thu Jul 16 13:30:18 2009 From: support at oeko.net (Toni Mueller) Date: Thu, 16 Jul 2009 11:30:18 +0200 Subject: 0.7.61: error_page 404 woes In-Reply-To: <20090713195734.GD15437@rambler-co.ru> References: <20090713170409.23831.qmail@oak.oeko.net> <20090713182225.GB15437@rambler-co.ru> <20090713184829.27178.qmail@oak.oeko.net> <20090713195734.GD15437@rambler-co.ru> Message-ID: <20090716093019.12254.qmail@oak.oeko.net> Hi, On Mon, 13.07.2009 at 23:57:34 +0400, Igor Sysoev wrote: > On Mon, Jul 13, 2009 at 08:48:29PM +0200, Toni Mueller wrote: > > On Mon, 13.07.2009 at 22:22:25 +0400, Igor Sysoev wrote: > > > On Mon, Jul 13, 2009 at 07:04:09PM +0200, Toni Mueller wrote: > > > > error_page 404 /; > > > > > > > > But this statement had no effect. I had to re-write the statement as > > > > > > > > error_page 404 http://$server_name/; > > > > > > > > to get the desired effect, but don't quite understand, why. > > > > > > Because the first statement does an internal redirect to /, > > > while the second one does an external redirect (302). > > > > thanks for the quick answer, but I don't quite understand: > > > > I already assumed that the second would cause an external redirect, but > > why the first statement would not lead to an internal rewrite of the > > URI and then try that through the remainder of the configuration, I > > don't see. > > What do you see in the first case ? You should see main page with > unchanged nonexistent URI in browser. in the first case, I see a standard 404 nginx error page. But I expected nginx to display the same thing that I see in the second case (I have a redirection for "/" to land somewhere else). Kind regards, --Toni++ From nginx-forum at nginx.us Thu Jul 16 13:41:05 2009 From: nginx-forum at nginx.us (davidc) Date: Thu, 16 Jul 2009 05:41:05 -0400 Subject: forum search Message-ID: <7d0491bc23ed0a9469f1529222278736.NginxMailingList@forum.nginx.org> This isn't specifically an nginx question but as the forum is served via nginx it could be. When I search the forum for "perl variable" it returns a '502 Bad Gateway' response and the version 'nginx/0.8.5'. This happens every time I try the search but not for other terms. Posted at Nginx Forum: http://forum.nginx.org/read.php?2,4102,4102#msg-4102 From support-nginx at oeko.net Thu Jul 16 13:55:38 2009 From: support-nginx at oeko.net (Toni Mueller) Date: Thu, 16 Jul 2009 11:55:38 +0200 Subject: Keep Alive piles up In-Reply-To: <20090629201857.GA99873@mdounin.ru> References: <20090629103818.GV66320@forum.nginx.org> <1ab82bc1f3e0954a2538288a4980621b.NginxMailingList@forum.nginx.org> <20090629144615.GC66320@forum.nginx.org> <20090629161004.GE66320@forum.nginx.org> <7b7119d7eebefa203cd21e74af6d9836.NginxMailingList@forum.nginx.org> <20090629182110.GF66320@forum.nginx.org> <1d8038ad189bdb81abb8eaea8438e35d.NginxMailingList@forum.nginx.org> <20090629201857.GA99873@mdounin.ru> Message-ID: <20090716095538.13093.qmail@oak.oeko.net> Hello! On Tue, 30.06.2009 at 00:18:57 +0400, Maxim Dounin wrote: > Looks like you assume that wiki == documentation. It's simply not > true. it seems that for those of us who can't read russian, it simply has to be true because we don't have much of a choice. > Yes, nested locations have drawbacks due to some inheritance bugs > (and hence they still aren't documented in official docs). But > if's are in fact nested locations under the hood - and they have I have trouble understanding how if()s can be locations under the hood, or how I could use variable testing (like eg. in geo-ip) without if()s. Kind regards, --Toni++ From nikon at puffy.pl Thu Jul 16 16:06:34 2009 From: nikon at puffy.pl (Tomasz Pajor) Date: Thu, 16 Jul 2009 14:06:34 +0200 Subject: fastcgi, try_files, problems Message-ID: <4A5F17CA.8040009@puffy.pl> Hello all, I have a problem with proper configuration of my page. What I want to achieve is that: - http://domain.com/ should point to an upstream bucket of fastcgi apps. - http://domain.com/administrator should be served by a fastcgi server on the same machine as balancer, so it should point to 127.0.0.1:3000 in this case - i think there is a way to simplify the two backend.php entries - all /js/ entries for domain.com should be served staticly - static.domain.com should be served staticly Don't know what am I doing wrong, but the debug says that when i go to http://domain.com/administrator it uses the location /administrator but then it's redirected to /index.php and it uses location /, so not as I intended. Can You guys help me sort this out? Configuration is as follows: upstream apps { server 10.0.1.50:3000; } server { listen 80; server_name domain.com; error_log /var/log/nginx/domain-error.log debug; location / { root /disk0/vhosts/domain/public; try_files $uri $uri /index.php; fastcgi_pass apps; #fastcgi_index index.php; fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name; include fastcgi_params; } location /administrator { root /disk0/vhosts/domain/public; try_files $uri $uri index.php; fastcgi_pass 127.0.0.1:3000; #fastcgi_index index.php; #fastcgi_param SCRIPT_FILENAME /disk0/vhosts/domain/public/index.php; fastcgi_param SCRIPT_FILENAME /disk0/vhosts/domain/public/$fastcgi_script_name; include fastcgi_params; } location /js/lib/xinha/plugins/ExtendedFileManager/backend.php { fastcgi_pass 127.0.0.1:3000; fastcgi_index backend.php; fastcgi_param SCRIPT_FILENAME /disk0/vhosts/domain/public/js/lib/xinha/plugins/ExtendedFileManager/backend.php; include fastcgi_params; access_log /var/log/nginx/backend1.log; } location /js/lib/xinha/plugins/ImageManager/backend.php { fastcgi_pass 127.0.0.1:3000; fastcgi_index backend.php; fastcgi_param SCRIPT_FILENAME /disk0/vhosts/domain/public/js/lib/xinha/plugins/ImageManager/backend.php; include fastcgi_params; access_log /var/log/nginx/backend2.log; } location ~ ^/(js)/ { root /disk0/vhosts/domain/public; expires 30d; access_log /var/log/nginx/js.log; } } server { listen 80; server_name static.domain.com; location / { root /disk0/vhosts/domain/static; access_log off; expires 30d; } } server { listen 80; server_name .domain2.com; location / { fastcgi_pass 127.0.0.1:3000; fastcgi_index index.php; fastcgi_param SCRIPT_FILENAME /disk0/vhosts/domain2/public/index.php; include fastcgi_params; } } From is at rambler-co.ru Thu Jul 16 16:31:41 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Thu, 16 Jul 2009 16:31:41 +0400 Subject: 0.7.61: error_page 404 woes In-Reply-To: <20090716093019.12254.qmail@oak.oeko.net> References: <20090713170409.23831.qmail@oak.oeko.net> <20090713182225.GB15437@rambler-co.ru> <20090713184829.27178.qmail@oak.oeko.net> <20090713195734.GD15437@rambler-co.ru> <20090716093019.12254.qmail@oak.oeko.net> Message-ID: <20090716123141.GE90866@rambler-co.ru> On Thu, Jul 16, 2009 at 11:30:18AM +0200, Toni Mueller wrote: > > Hi, > > On Mon, 13.07.2009 at 23:57:34 +0400, Igor Sysoev wrote: > > On Mon, Jul 13, 2009 at 08:48:29PM +0200, Toni Mueller wrote: > > > On Mon, 13.07.2009 at 22:22:25 +0400, Igor Sysoev wrote: > > > > On Mon, Jul 13, 2009 at 07:04:09PM +0200, Toni Mueller wrote: > > > > > error_page 404 /; > > > > > > > > > > But this statement had no effect. I had to re-write the statement as > > > > > > > > > > error_page 404 http://$server_name/; > > > > > > > > > > to get the desired effect, but don't quite understand, why. > > > > > > > > Because the first statement does an internal redirect to /, > > > > while the second one does an external redirect (302). > > > > > > thanks for the quick answer, but I don't quite understand: > > > > > > I already assumed that the second would cause an external redirect, but > > > why the first statement would not lead to an internal rewrite of the > > > URI and then try that through the remainder of the configuration, I > > > don't see. > > > > What do you see in the first case ? You should see main page with > > unchanged nonexistent URI in browser. > > in the first case, I see a standard 404 nginx error page. But I > expected nginx to display the same thing that I see in the second case > (I have a redirection for "/" to land somewhere else). Could you show your configuration ? -- Igor Sysoev http://sysoev.ru/en/ From is at rambler-co.ru Thu Jul 16 16:37:04 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Thu, 16 Jul 2009 16:37:04 +0400 Subject: Maxmind GeoIP City Module In-Reply-To: <4A5336CD.9050308@theactivitypeople.co.uk> References: <4A5336CD.9050308@theactivitypeople.co.uk> Message-ID: <20090716123704.GF90866@rambler-co.ru> On Tue, Jul 07, 2009 at 12:51:41PM +0100, Phillip Oldham wrote: > We're in need of a GeoIP module which works with the current development > branch of nginx and can resolve an IP to City-level using Maxmind's > GeoIP Database (the .dat files). We use nginx extensively but don't have > experience in C to develop such a module in-house. Are there any module > developers on the list who would be interested in such a project? > > We're flexible regarding the funding of development for such a module, > and also flexible on whether the finshed module is open-sourced and > released back to the community. > > If anyone is interested please feel free to mail me off-list with any > thoughts/costs/etc. The module is in the attached patch. ./configure ---with-http_geoip_module nginx.conf: http { geoip_country path/to/GeoIP.dat; geoip_city path/to/GeoLiteCity.dat; You may use only one base or both bases. Varaibles available with geoip_country base: $geoip_country_code $geoip_country_code3 $geoip_country_name Varaibles available with geoip_city base: $geoip_city_country_code $geoip_city_country_code3 $geoip_city_country_name $geoip_region $geoip_city $geoip_postal_code -- Igor Sysoev http://sysoev.ru/en/ -------------- next part -------------- Index: src/http/modules/ngx_http_geoip_module.c =================================================================== --- src/http/modules/ngx_http_geoip_module.c (revision 0) +++ src/http/modules/ngx_http_geoip_module.c (revision 0) @@ -0,0 +1,360 @@ + +/* + * Copyright (C) Igor Sysoev + */ + + +#include +#include +#include + +#include +#include + + +typedef struct { + GeoIP *country; + GeoIP *city; +} ngx_http_geoip_conf_t; + + +typedef struct { + ngx_str_t *name; + uintptr_t data; +} ngx_http_geoip_var_t; + + +typedef const char *(*ngx_http_geoip_variable_handler_pt)(GeoIP *, u_long addr); + +static ngx_int_t ngx_http_geoip_country_variable(ngx_http_request_t *r, + ngx_http_variable_value_t *v, uintptr_t data); +static ngx_int_t ngx_http_geoip_city_variable(ngx_http_request_t *r, + ngx_http_variable_value_t *v, uintptr_t data); + +static ngx_int_t ngx_http_geoip_add_variables(ngx_conf_t *cf); +static void *ngx_http_geoip_create_conf(ngx_conf_t *cf); +static char *ngx_http_geoip_country(ngx_conf_t *cf, ngx_command_t *cmd, + void *conf); +static char *ngx_http_geoip_city(ngx_conf_t *cf, ngx_command_t *cmd, + void *conf); +static void ngx_http_geoip_cleanup(void *data); + + +static ngx_command_t ngx_http_geoip_commands[] = { + + { ngx_string("geoip_country"), + NGX_HTTP_MAIN_CONF|NGX_CONF_TAKE1, + ngx_http_geoip_country, + NGX_HTTP_MAIN_CONF_OFFSET, + 0, + NULL }, + + { ngx_string("geoip_city"), + NGX_HTTP_MAIN_CONF|NGX_CONF_TAKE1, + ngx_http_geoip_city, + NGX_HTTP_MAIN_CONF_OFFSET, + 0, + NULL }, + + ngx_null_command +}; + + +static ngx_http_module_t ngx_http_geoip_module_ctx = { + ngx_http_geoip_add_variables, /* preconfiguration */ + NULL, /* postconfiguration */ + + ngx_http_geoip_create_conf, /* create main configuration */ + NULL, /* init main configuration */ + + NULL, /* create server configuration */ + NULL, /* merge server configuration */ + + NULL, /* create location configuration */ + NULL /* merge location configuration */ +}; + + +ngx_module_t ngx_http_geoip_module = { + NGX_MODULE_V1, + &ngx_http_geoip_module_ctx, /* module context */ + ngx_http_geoip_commands, /* module directives */ + NGX_HTTP_MODULE, /* module type */ + NULL, /* init master */ + NULL, /* init module */ + NULL, /* init process */ + NULL, /* init thread */ + NULL, /* exit thread */ + NULL, /* exit process */ + NULL, /* exit master */ + NGX_MODULE_V1_PADDING +}; + + +static ngx_http_variable_t ngx_http_geoip_vars[] = { + + { ngx_string("geoip_country_code"), NULL, ngx_http_geoip_country_variable, + (uintptr_t) GeoIP_country_code_by_ipnum, 0, 0 }, + + { ngx_string("geoip_country_code3"), NULL, ngx_http_geoip_country_variable, + (uintptr_t) GeoIP_country_code3_by_ipnum, 0, 0 }, + + { ngx_string("geoip_country_name"), NULL, ngx_http_geoip_country_variable, + (uintptr_t) GeoIP_country_name_by_ipnum, 0, 0 }, + + { ngx_string("geoip_city_country_code"), NULL, ngx_http_geoip_city_variable, + offsetof(GeoIPRecord, country_code), 0, 0 }, + + { ngx_string("geoip_city_country_code3"), NULL, + ngx_http_geoip_city_variable, + offsetof(GeoIPRecord, country_code3), 0, 0 }, + + { ngx_string("geoip_city_country_name"), NULL, ngx_http_geoip_city_variable, + offsetof(GeoIPRecord, country_name), 0, 0 }, + + { ngx_string("geoip_region"), NULL, + ngx_http_geoip_city_variable, + offsetof(GeoIPRecord, region), 0, 0 }, + + { ngx_string("geoip_city"), NULL, + ngx_http_geoip_city_variable, + offsetof(GeoIPRecord, city), 0, 0 }, + + { ngx_string("geoip_postal_code"), NULL, + ngx_http_geoip_city_variable, + offsetof(GeoIPRecord, postal_code), 0, 0 }, + + { ngx_null_string, NULL, NULL, 0, 0, 0 } +}; + + +static ngx_int_t +ngx_http_geoip_country_variable(ngx_http_request_t *r, + ngx_http_variable_value_t *v, uintptr_t data) +{ + ngx_http_geoip_variable_handler_pt handler = + (ngx_http_geoip_variable_handler_pt) data; + + u_long addr; + const char *val; + struct sockaddr_in *sin; + ngx_http_geoip_conf_t *gcf; + + gcf = ngx_http_get_module_main_conf(r, ngx_http_geoip_module); + + if (gcf->country == NULL) { + goto not_found; + } + + if (r->connection->sockaddr->sa_family != AF_INET) { + goto not_found; + } + + sin = (struct sockaddr_in *) r->connection->sockaddr; + addr = ntohl(sin->sin_addr.s_addr); + + val = handler(gcf->country, addr); + + if (val == NULL) { + goto not_found; + } + + v->len = ngx_strlen(val); + v->valid = 1; + v->no_cacheable = 0; + v->not_found = 0; + v->data = (u_char *) val; + + return NGX_OK; + +not_found: + + v->not_found = 1; + + return NGX_OK; +} + + +static ngx_int_t +ngx_http_geoip_city_variable(ngx_http_request_t *r, + ngx_http_variable_value_t *v, uintptr_t data) +{ + u_long addr; + char *val; + GeoIPRecord *gr; + struct sockaddr_in *sin; + ngx_http_geoip_conf_t *gcf; + + gcf = ngx_http_get_module_main_conf(r, ngx_http_geoip_module); + + if (gcf->city == NULL) { + goto not_found; + } + + if (r->connection->sockaddr->sa_family != AF_INET) { + goto not_found; + } + + sin = (struct sockaddr_in *) r->connection->sockaddr; + addr = ntohl(sin->sin_addr.s_addr); + + gr = GeoIP_record_by_ipnum(gcf->city, addr); + + if (gr == NULL) { + goto not_found; + } + + val = *(char **) ((char *) gr + data); + + if (val == NULL) { + goto not_found; + } + + v->len = ngx_strlen(val); + v->valid = 1; + v->no_cacheable = 0; + v->not_found = 0; + v->data = (u_char *) val; + + return NGX_OK; + +not_found: + + v->not_found = 1; + + return NGX_OK; +} + + +static ngx_int_t +ngx_http_geoip_add_variables(ngx_conf_t *cf) +{ + ngx_http_variable_t *var, *v; + + for (v = ngx_http_geoip_vars; v->name.len; v++) { + var = ngx_http_add_variable(cf, &v->name, v->flags); + if (var == NULL) { + return NGX_ERROR; + } + + var->get_handler = v->get_handler; + var->data = v->data; + } + + return NGX_OK; +} + + +static void * +ngx_http_geoip_create_conf(ngx_conf_t *cf) +{ + ngx_pool_cleanup_t *cln; + ngx_http_geoip_conf_t *conf; + + conf = ngx_pcalloc(cf->pool, sizeof(ngx_http_geoip_conf_t)); + if (conf == NULL) { + return NULL; + } + + cln = ngx_pool_cleanup_add(cf->pool, 0); + if (cln == NULL) { + return NULL; + } + + cln->handler = ngx_http_geoip_cleanup; + cln->data = conf; + + return conf; +} + + +static char * +ngx_http_geoip_country(ngx_conf_t *cf, ngx_command_t *cmd, void *conf) +{ + ngx_http_geoip_conf_t *gcf = conf; + + ngx_str_t *value; + + if (gcf->country) { + return "is duplicate"; + } + + value = cf->args->elts; + + gcf->country = GeoIP_open((char *) value[1].data, GEOIP_MEMORY_CACHE); + + if (gcf->country == NULL) { + ngx_conf_log_error(NGX_LOG_EMERG, cf, 0, + "GeoIO_open(\"%V\") failed", &value[1]); + + return NGX_CONF_ERROR; + } + + switch (gcf->country->databaseType) { + + case GEOIP_COUNTRY_EDITION: + case GEOIP_PROXY_EDITION: + case GEOIP_NETSPEED_EDITION: + + return NGX_CONF_OK; + + default: + ngx_conf_log_error(NGX_LOG_EMERG, cf, 0, + "invalid GeoIP database \"%V\" type:%d", + &value[1], gcf->country->databaseType); + return NGX_CONF_ERROR; + } +} + + +static char * +ngx_http_geoip_city(ngx_conf_t *cf, ngx_command_t *cmd, void *conf) +{ + ngx_http_geoip_conf_t *gcf = conf; + + ngx_str_t *value; + + if (gcf->city) { + return "is duplicate"; + } + + value = cf->args->elts; + + gcf->city = GeoIP_open((char *) value[1].data, GEOIP_MEMORY_CACHE); + + if (gcf->city == NULL) { + ngx_conf_log_error(NGX_LOG_EMERG, cf, 0, + "GeoIO_open(\"%V\") failed", &value[1]); + + return NGX_CONF_ERROR; + } + + switch (gcf->city->databaseType) { + + case GEOIP_CITY_EDITION_REV0: + case GEOIP_CITY_EDITION_REV1: + + return NGX_CONF_OK; + + default: + ngx_conf_log_error(NGX_LOG_EMERG, cf, 0, + "invalid GeoIP City database \"%V\" type:%d", + &value[1], gcf->city->databaseType); + return NGX_CONF_ERROR; + } +} + + +static void +ngx_http_geoip_cleanup(void *data) +{ + ngx_http_geoip_conf_t *gcf = data; + + if (gcf->country) { + GeoIP_delete(gcf->country); + } + + if (gcf->city) { + GeoIP_delete(gcf->city); + } +} Index: auto/sources =================================================================== --- auto/sources (revision 2303) +++ auto/sources (working copy) @@ -377,6 +377,10 @@ HTTP_GEO_SRCS=src/http/modules/ngx_http_geo_module.c +HTTP_GEOIP_MODULE=ngx_http_geoip_module +HTTP_GEOIP_SRCS=src/http/modules/ngx_http_geoip_module.c + + HTTP_MAP_MODULE=ngx_http_map_module HTTP_MAP_SRCS=src/http/modules/ngx_http_map_module.c Index: auto/options =================================================================== --- auto/options (revision 2303) +++ auto/options (working copy) @@ -71,6 +71,7 @@ HTTP_RANDOM_INDEX=NO HTTP_STATUS=NO HTTP_GEO=YES +HTTP_GEOIP=NO HTTP_MAP=YES HTTP_REFERER=YES HTTP_REWRITE=YES @@ -184,6 +185,7 @@ --with-http_addition_module) HTTP_ADDITION=YES ;; --with-http_xslt_module) HTTP_XSLT=YES ;; --with-http_image_filter_module) HTTP_IMAGE_FILTER=YES ;; + --with-http_geoip_module) HTTP_GEOIP=YES ;; --with-http_sub_module) HTTP_SUB=YES ;; --with-http_dav_module) HTTP_DAV=YES ;; --with-http_flv_module) HTTP_FLV=YES ;; @@ -310,6 +312,7 @@ --with-http_addition_module enable ngx_http_addition_module --with-http_xslt_module enable ngx_http_xslt_module --with-http_image_filter_module enable ngx_http_image_filter_module + --with-http_geoip_module enable ngx_http_geoip_link_module --with-http_sub_module enable ngx_http_sub_module --with-http_dav_module enable ngx_http_dav_module --with-http_flv_module enable ngx_http_flv_module Index: auto/lib/geoip/conf =================================================================== --- auto/lib/geoip/conf (revision 0) +++ auto/lib/geoip/conf (revision 0) @@ -0,0 +1,78 @@ + +# Copyright (C) Igor Sysoev + + + ngx_feature="GeoIP library" + ngx_feature_name= + ngx_feature_run=no + ngx_feature_incs= + ngx_feature_path= + ngx_feature_libs="-lGeoIP" + ngx_feature_test="GeoIP_open(NULL, 0)" + . auto/feature + + +if [ $ngx_found = no ]; then + + # FreeBSD port + + ngx_feature="GeoIP library in /usr/local/" + + if [ $NGX_RPATH = YES ]; then + ngx_feature_libs="-R/usr/local/lib -L/usr/local/lib -lGeoIP" + else + ngx_feature_libs="-L/usr/local/lib -lGeoIP" + fi + + . auto/feature +fi + + +if [ $ngx_found = no ]; then + + # NetBSD port + + ngx_feature="GeoIP library in /usr/pkg/" + ngx_feature_path="/usr/pkg/include/" + + if [ $NGX_RPATH = YES ]; then + ngx_feature_libs="-R/usr/pkg/lib -L/usr/pkg/lib -lGeoIP" + else + ngx_feature_libs="-L/usr/pkg/lib -lGeoIP" + fi + + . auto/feature +fi + + +if [ $ngx_found = no ]; then + + # MacPorts + + ngx_feature="GeoIP library in /opt/local/" + ngx_feature_path="/opt/local/include" + + if [ $NGX_RPATH = YES ]; then + ngx_feature_libs="-R/opt/local/lib -L/opt/local/lib -lGeoIP" + else + ngx_feature_libs="-L/opt/local/lib -lGeoIP" + fi + + . auto/feature +fi + + +if [ $ngx_found = yes ]; then + CORE_LIBS="$CORE_LIBS $ngx_feature_libs" + +else + +cat << END + +$0: error: the GeoIP module requires the GeoIP library. +You can either do not enable the module or install the library. + +END + + exit 1 +fi Index: auto/lib/conf =================================================================== --- auto/lib/conf (revision 2303) +++ auto/lib/conf (working copy) @@ -67,6 +67,9 @@ . auto/lib/perl/conf fi +if [ $HTTP_GEOIP = YES ]; then + . auto/lib/geoip/conf +fi if [ $NGX_GOOGLE_PERFTOOLS = YES ]; then . auto/lib/google-perftools/conf fi Index: auto/modules =================================================================== --- auto/modules (revision 2303) +++ auto/modules (working copy) @@ -240,6 +240,12 @@ HTTP_SRCS="$HTTP_SRCS $HTTP_GEO_SRCS" fi +if [ $HTTP_GEOIP = YES ]; then + have=NGX_HTTP_GEOIP . auto/have + HTTP_MODULES="$HTTP_MODULES $HTTP_GEOIP_MODULE" + HTTP_SRCS="$HTTP_SRCS $HTTP_GEOIP_SRCS" +fi + if [ $HTTP_MAP = YES ]; then have=NGX_HTTP_MAP . auto/have HTTP_MODULES="$HTTP_MODULES $HTTP_MAP_MODULE" From mdounin at mdounin.ru Thu Jul 16 16:41:59 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Thu, 16 Jul 2009 16:41:59 +0400 Subject: fastcgi, try_files, problems In-Reply-To: <4A5F17CA.8040009@puffy.pl> References: <4A5F17CA.8040009@puffy.pl> Message-ID: <20090716124159.GI1050@mdounin.ru> Hello! On Thu, Jul 16, 2009 at 02:06:34PM +0200, Tomasz Pajor wrote: > Hello all, > > I have a problem with proper configuration of my page. > > What I want to achieve is that: > - http://domain.com/ should point to an upstream bucket of fastcgi apps. > - http://domain.com/administrator should be served by a fastcgi server > on the same machine as balancer, so it should point to 127.0.0.1:3000 in > this case > - i think there is a way to simplify the two backend.php entries > - all /js/ entries for domain.com should be served staticly > - static.domain.com should be served staticly > > Don't know what am I doing wrong, but the debug says that when i go to > http://domain.com/administrator it uses the location /administrator but > then it's redirected to /index.php and it uses location /, so not as I > intended. > > Can You guys help me sort this out? > > Configuration is as follows: > > upstream apps { > server 10.0.1.50:3000; > } > > server { > listen 80; > server_name domain.com; > error_log /var/log/nginx/domain-error.log debug; > > location / { > root /disk0/vhosts/domain/public; > try_files $uri $uri /index.php; See no reasons to test $uri twice. > fastcgi_pass apps; > #fastcgi_index index.php; > fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name; > include fastcgi_params; > } > > location /administrator { > root /disk0/vhosts/domain/public; > try_files $uri $uri index.php; You fallback to 'index.php' which is wrong as far as I see (it produces redirect to 'index.php' in case of no file found - it's simply invalid since has no leading '/'). Probably you mean to use try_files $uri /administrator/index.php; instead, and probably it's a good idea to specify separate location for /administrator/index.php to avoid internal redirection loop in case there will be no /administrator/index.php file. > fastcgi_pass 127.0.0.1:3000; > #fastcgi_index index.php; > #fastcgi_param SCRIPT_FILENAME > /disk0/vhosts/domain/public/index.php; > fastcgi_param SCRIPT_FILENAME > /disk0/vhosts/domain/public/$fastcgi_script_name; > include fastcgi_params; > } > > location /js/lib/xinha/plugins/ExtendedFileManager/backend.php { > fastcgi_pass 127.0.0.1:3000; > fastcgi_index backend.php; > fastcgi_param SCRIPT_FILENAME > /disk0/vhosts/domain/public/js/lib/xinha/plugins/ExtendedFileManager/backend.php; > include fastcgi_params; > access_log /var/log/nginx/backend1.log; > } > > location /js/lib/xinha/plugins/ImageManager/backend.php { > fastcgi_pass 127.0.0.1:3000; > fastcgi_index backend.php; > fastcgi_param SCRIPT_FILENAME > /disk0/vhosts/domain/public/js/lib/xinha/plugins/ImageManager/backend.php; > include fastcgi_params; > access_log /var/log/nginx/backend2.log; > } > > location ~ ^/(js)/ { Any reasons to use regex here? It should be enough to use normal location /js/ { unless you want to render the above '/js/.../backend.php' locations useless. Maxim Dounin > root /disk0/vhosts/domain/public; > expires 30d; > access_log /var/log/nginx/js.log; > } > } > > server { > listen 80; > server_name static.domain.com; > > location / { > root /disk0/vhosts/domain/static; > access_log off; > expires 30d; > } > } > > server { > listen 80; > server_name .domain2.com; > > location / { > fastcgi_pass 127.0.0.1:3000; > fastcgi_index index.php; > fastcgi_param SCRIPT_FILENAME > /disk0/vhosts/domain2/public/index.php; > include fastcgi_params; > } > } > From dick at fouter.net Thu Jul 16 16:48:27 2009 From: dick at fouter.net (Dick Middleton) Date: Thu, 16 Jul 2009 13:48:27 +0100 Subject: fastcgi, try_files, problems In-Reply-To: <4A5F17CA.8040009@puffy.pl> References: <4A5F17CA.8040009@puffy.pl> Message-ID: <4A5F219B.5040504@fouter.net> Tomasz Pajor wrote: > I have a problem with proper configuration of my page. Coo! Let me have a go. I'm brand new to nginx so just testing my reading ability :) > Don't know what am I doing wrong, but the debug says that when i go to > http://domain.com/administrator it uses the location /administrator but > then it's redirected to /index.php and it uses location /, so not as I > intended. Is it the order of location directives: location / satisfies /administrator before location /administrator So change the order in config - am I right? Dick From maccaday at gmail.com Thu Jul 16 16:52:43 2009 From: maccaday at gmail.com (Marcus Clyne) Date: Thu, 16 Jul 2009 15:52:43 +0300 Subject: Daemon directory string In-Reply-To: <20090716012116.GD1050@mdounin.ru> References: <57B78091-3ED0-4DBC-988F-E64A4C01F212@tinyapps.org> <664722396.20090715200227@csdoc.com> <43286079-5C19-4249-8DA8-824434B25AE6@tinyapps.org> <4A5E2A56.4070006@gmail.com> <20090715193756.GC1050@mdounin.ru> <4A5E6568.7000307@gmail.com> <20090716012116.GD1050@mdounin.ru> Message-ID: <4A5F229B.7080505@gmail.com> Maxim Dounin wrote: > Hello! > > On Thu, Jul 16, 2009 at 02:25:28AM +0300, Marcus Clyne wrote: > > >> Hi, >> >> What's the best way to get the path of the directory that nginx was >> called from (as part of module development)? Is this stored in a string >> somewhere? >> > > No. You should use ngx_conf_full_name() to resolve paths to > absolute ones. It uses cycle->prefix (or cycle->conf_prefix), and > this in turn is set to startup directory if there is no prefix > (either compiled in or specified via -p switch). > > Maxim Dounin > > Thanks. From is at rambler-co.ru Thu Jul 16 16:51:35 2009 From: is at rambler-co.ru (Igor Sysoev) Date: Thu, 16 Jul 2009 16:51:35 +0400 Subject: problem w/ gzip when hiding Content-Type header from upstream server In-Reply-To: <925f8f7428ee278a6300710bc88646ab.NginxMailingList@forum.nginx.org> References: <69e0c4acac418c641823f567866f9e3c.NginxMailingList@forum.nginx.org> <925f8f7428ee278a6300710bc88646ab.NginxMailingList@forum.nginx.org> Message-ID: <20090716125135.GI90866@rambler-co.ru> On Thu, Jul 16, 2009 at 04:55:00AM -0400, bavery wrote: > I did some more checking, and am now finding that other .js & .css files from the upstream server are not being compressed as well, except that these files simply are served from the upstream server with no content-type header at all. > > So why does nginx not recognize the file type it is handling based on mime-type configuration when dealing with proxied files? nginx think that upstream should set Content-Type by itself. Probably I need to extend "add_header" to set content type: location /combined.css { proxy_pass .. proxy_hide_header Content-Type; add_header Content-Type text/css; or implement "override_type" to set content-type based on mime.types file or just string: location /combined.css { proxy_pass .. override_type [on|off|mime-type]; -- Igor Sysoev http://sysoev.ru/en/ From nikon at puffy.pl Thu Jul 16 17:07:29 2009 From: nikon at puffy.pl (Tomasz Pajor) Date: Thu, 16 Jul 2009 15:07:29 +0200 Subject: fastcgi, try_files, problems In-Reply-To: <20090716124159.GI1050@mdounin.ru> References: <4A5F17CA.8040009@puffy.pl> <20090716124159.GI1050@mdounin.ru> Message-ID: <4A5F2611.8020009@puffy.pl> Maxim Dounin wrote: > Hello! > > On Thu, Jul 16, 2009 at 02:06:34PM +0200, Tomasz Pajor wrote: > > >> Hello all, >> >> I have a problem with proper configuration of my page. >> >> What I want to achieve is that: >> - http://domain.com/ should point to an upstream bucket of fastcgi apps. >> - http://domain.com/administrator should be served by a fastcgi server >> on the same machine as balancer, so it should point to 127.0.0.1:3000 in >> this case >> - i think there is a way to simplify the two backend.php entries >> - all /js/ entries for domain.com should be served staticly >> - static.domain.com should be served staticly >> >> Don't know what am I doing wrong, but the debug says that when i go to >> http://domain.com/administrator it uses the location /administrator but >> then it's redirected to /index.php and it uses location /, so not as I >> intended. >> >> Can You guys help me sort this out? >> >> Configuration is as follows: >> >> upstream apps { >> server 10.0.1.50:3000; >> } >> >> server { >> listen 80; >> server_name domain.com; >> error_log /var/log/nginx/domain-error.log debug; >> >> location / { >> root /disk0/vhosts/domain/public; >> try_files $uri $uri /index.php; >> > > See no reasons to test $uri twice. > > >> fastcgi_pass apps; >> #fastcgi_index index.php; >> fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name; >> include fastcgi_params; >> } >> >> location /administrator { >> root /disk0/vhosts/domain/public; >> try_files $uri $uri index.php; >> > > You fallback to 'index.php' which is wrong as far as I see (it > produces redirect to 'index.php' in case of no file found - it's > simply invalid since has no leading '/'). Probably you mean to > use > > try_files $uri /administrator/index.php; > > instead, and probably it's a good idea to specify separate > location for /administrator/index.php to avoid internal > redirection loop in case there will be no /administrator/index.php > file. > There is no /administrator/index.php file. I want to simply separate the administrator backend, and serve it from the same machine as balancer, because i've got all the gallery photos on this machine. I'm migrating this from apache. .htaccess says: RewriteEngine On RewriteCond %{REQUEST_FILENAME} -s [OR] RewriteCond %{REQUEST_FILENAME} -l [OR] RewriteCond %{REQUEST_FILENAME} -d RewriteRule ^.*$ - [NC,L] RewriteRule ^.*$ index.php [NC,L] >> fastcgi_pass 127.0.0.1:3000; >> #fastcgi_index index.php; >> #fastcgi_param SCRIPT_FILENAME >> /disk0/vhosts/domain/public/index.php; >> fastcgi_param SCRIPT_FILENAME >> /disk0/vhosts/domain/public/$fastcgi_script_name; >> include fastcgi_params; >> } >> >> location /js/lib/xinha/plugins/ExtendedFileManager/backend.php { >> fastcgi_pass 127.0.0.1:3000; >> fastcgi_index backend.php; >> fastcgi_param SCRIPT_FILENAME >> /disk0/vhosts/domain/public/js/lib/xinha/plugins/ExtendedFileManager/backend.php; >> include fastcgi_params; >> access_log /var/log/nginx/backend1.log; >> } >> >> location /js/lib/xinha/plugins/ImageManager/backend.php { >> fastcgi_pass 127.0.0.1:3000; >> fastcgi_index backend.php; >> fastcgi_param SCRIPT_FILENAME >> /disk0/vhosts/domain/public/js/lib/xinha/plugins/ImageManager/backend.php; >> include fastcgi_params; >> access_log /var/log/nginx/backend2.log; >> } >> >> location ~ ^/(js)/ { >> > > Any reasons to use regex here? It should be enough to use normal > > location /js/ { > > unless you want to render the above '/js/.../backend.php' locations > useless. > Yes the regex is useless in this case. Is there any way of merging this to */backend.php locations? From mdounin at mdounin.ru Thu Jul 16 17:18:23 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Thu, 16 Jul 2009 17:18:23 +0400 Subject: fastcgi, try_files, problems In-Reply-To: <4A5F219B.5040504@fouter.net> References: <4A5F17CA.8040009@puffy.pl> <4A5F219B.5040504@fouter.net> Message-ID: <20090716131822.GL1050@mdounin.ru> Hello! On Thu, Jul 16, 2009 at 01:48:27PM +0100, Dick Middleton wrote: > Tomasz Pajor wrote: > > > I have a problem with proper configuration of my page. > > Coo! Let me have a go. I'm brand new to nginx so just testing my > reading ability :) > >> Don't know what am I doing wrong, but the debug says that when i go to >> http://domain.com/administrator it uses the location /administrator but >> then it's redirected to /index.php and it uses location /, so not as I >> intended. > > Is it the order of location directives: location / satisfies > /administrator before location /administrator > > So change the order in config - am I right? No. For normal location (not regex) order doesn't matter, nginx always uses most specific (i.e. longest match). See http://wiki.nginx.org/NginxHttpCoreModule#location for details. Maxim Dounin From grails at jmsd.co.uk Thu Jul 16 17:33:43 2009 From: grails at jmsd.co.uk (John Moore) Date: Thu, 16 Jul 2009 14:33:43 +0100 Subject: Simple permissions problem Message-ID: <4A5F2C37.50101@jmsd.co.uk> Sorry, this should be dead simple but it's not working for me... I have a directory, called 'php', within my main directory which contains fragments of HTML to be included in my main pages (either by SSI or by PHP includes). I do not want any of these fragments to be directly accessible by a browser. That is, I want 'http://www.mysite.com/php/fragment.html' to fail. I have the following in my nginx.conf, but it doesn't work - 'http://www.mysite.com/php/fragment.html' still brings up the fragment in the browser. What do I need to do? (And, on a completely separate note, do I have to specify the full path to the root in each of my hosts, or is there some shortcut for this)? server { server_name www.mysite.com; listen 80; location /php { root /var/www/nginx-default/www.mysite.com; deny all; } } JM From mdounin at mdounin.ru Thu Jul 16 20:20:29 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Thu, 16 Jul 2009 20:20:29 +0400 Subject: Keep Alive piles up In-Reply-To: <20090716095538.13093.qmail@oak.oeko.net> References: <1ab82bc1f3e0954a2538288a4980621b.NginxMailingList@forum.nginx.org> <20090629144615.GC66320@forum.nginx.org> <20090629161004.GE66320@forum.nginx.org> <7b7119d7eebefa203cd21e74af6d9836.NginxMailingList@forum.nginx.org> <20090629182110.GF66320@forum.nginx.org> <1d8038ad189bdb81abb8eaea8438e35d.NginxMailingList@forum.nginx.org> <20090629201857.GA99873@mdounin.ru> <20090716095538.13093.qmail@oak.oeko.net> Message-ID: <20090716162029.GO1050@mdounin.ru> Hello! On Thu, Jul 16, 2009 at 11:55:38AM +0200, Toni Mueller wrote: > > Hello! > > On Tue, 30.06.2009 at 00:18:57 +0400, Maxim Dounin wrote: > > Looks like you assume that wiki == documentation. It's simply not > > true. > > it seems that for those of us who can't read russian, it simply has to > be true because we don't have much of a choice. It's probably a good idea to maintain translation of documentation somewhere, may be just change wiki policy for generic pages. For now, feel free to use google translate or yahoo babelfish. > > Yes, nested locations have drawbacks due to some inheritance bugs > > (and hence they still aren't documented in official docs). But > > if's are in fact nested locations under the hood - and they have > > I have trouble understanding how if()s can be locations under the hood, > or how I could use variable testing (like eg. in geo-ip) without if()s. Looks like statements for me, not questions. Anyway, here is some explanations: 1. Directive if() used inside location creates another location configuration object via ngx_http_add_location() (and then it's used in case if() condition passes), see src/http/modules/ngx_http_rewrite_module.c:ngx_http_rewrite_if() for details. 2. In some cases you can avoid using if()'s. In some you can't. If you can - in most cases it's better choice to avoid. If you can't - it's somethimes better to avoid the whole case then to use if(). Normally if's will work. But somethimes they produce unexpected results or even coredumps. The bad thing is that previously good configuration that uses if's may be rendered invalid just by adding some minor unrelated directives (or vice versa, good configuration without if's may be rendered invalid by adding unrelated if's). Here are some examples: 1. Good, try_files working as expected: location /try-files { try_files /file @fallback; } Bad, try_files not working at all: location /if-try-files { try_files /file @fallback; set $true 1; if ($true) { # nothing } } 2. Good, works as expected: location /no-crash { set $true 1; if ($true) { # fastcgi_pass here fastcgi_pass 127.0.0.1:9000; } } Bad, crashes: location /crash { set $true 1; if ($true) { # fastcgi_pass here fastcgi_pass 127.0.0.1:9000; } if ($true) { # no handler here } } The other bad thing - it's not clear how to fix this without breaking compatibility. Note that if() used inside server{} block and if() used inside location block are quite a different beasts. While the former is limited to safe things and mostly harmless - latter is evil. On the other hand, using server-scope if's is just inefficient if you need them only for some specific location. Maxim Dounin From mike503 at gmail.com Thu Jul 16 20:42:53 2009 From: mike503 at gmail.com (Michael Shadle) Date: Thu, 16 Jul 2009 09:42:53 -0700 Subject: fastcgi, try_files, problems In-Reply-To: <20090716124159.GI1050@mdounin.ru> References: <4A5F17CA.8040009@puffy.pl> <20090716124159.GI1050@mdounin.ru> Message-ID: On Thu, Jul 16, 2009 at 5:41 AM, Maxim Dounin wrote: > See no reasons to test $uri twice. I think it's a typo. probably meant try_files $uri $uri/ /index.php; From mike503 at gmail.com Thu Jul 16 20:46:08 2009 From: mike503 at gmail.com (Michael Shadle) Date: Thu, 16 Jul 2009 09:46:08 -0700 Subject: FastCGI application not receiving multi-part data In-Reply-To: <14a2ed3e7b6b55d3caacbc706609cefb.NginxMailingList@forum.nginx.org> References: <6474272efa8edf4f503a18f5fe75a2cf.NginxMailingList@forum.nginx.org> <14a2ed3e7b6b55d3caacbc706609cefb.NginxMailingList@forum.nginx.org> Message-ID: post your configuration On Thu, Jul 16, 2009 at 2:27 AM, gerryw wrote: > Hello All, > > I also did a test with just the vanilla FastCGI API calls. The only env related variable I was able to get was "REQUEST_METHOD". None of the rest were set. > > Thanks, > Gerry > > Posted at Nginx Forum: http://forum.nginx.org/read.php?2,4091,4101#msg-4101 > > > From mdounin at mdounin.ru Thu Jul 16 21:58:56 2009 From: mdounin at mdounin.ru (Maxim Dounin) Date: Thu, 16 Jul 2009 21:58:56 +0400 Subject: fastcgi, try_files, problems In-Reply-To: <4A5F2611.8020009@puffy.pl> References: <4A5F17CA.8040009@puffy.pl> <20090716124159.GI1050@mdounin.ru> <4A5F2611.8020009@puffy.pl> Message-ID: <20090716175856.GS1050@mdounin.ru> Hello! On Thu, Jul 16, 2009 at 03:07:29PM +0200, Tomasz Pajor wrote: > Maxim Dounin wrote: >> Hello! >> >> On Thu, Jul 16, 2009 at 02:06:34PM +0200, Tomasz Pajor wrote: >> >> >>> Hello all, >>> >>> I have a problem with proper configuration of my page. >>> >>> What I want to achieve is that: >>> - http://domain.com/ should point to an upstream bucket of fastcgi apps. >>> - http://domain.com/administrator should be served by a fastcgi >>> server on the same machine as balancer, so it should point to >>> 127.0.0.1:3000 in this case >>> - i think there is a way to simplify the two backend.php entries >>> - all /js/ entries for domain.com should be served staticly >>> - static.domain.com should be served staticly >>> >>> Don't know what am I doing wrong, but the debug says that when i go >>> to http://domain.com/administrator it uses the location >>> /administrator but then it's redirected to /index.php and it uses >>> location /, so not as I intended. >>> >>> Can You guys help me sort this out? >>> >>> Configuration is as follows: >>> >>> upstream apps { >>> server 10.0.1.50:3000; >>> } >>> >>> server { >>> listen 80; >>> server_name domain.com; >>> error_log /var/log/nginx/domain-error.log debug; >>> >>> location / { >>> root /disk0/vhosts/domain/public; >>> try_files $uri $uri /index.php; >>> >> >> See no reasons to test $uri twice. >> >> >>> fastcgi_pass apps; >>> #fastcgi_index index.php; >>> fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name; >>> include fastcgi_params; >>> } >>> >>> location /administrator { >>> root /disk0/vhosts/domain/public; >>> try_files $uri $uri index.php; >>> >> >> You fallback to 'index.php' which is wrong as far as I see (it >> produces redirect to 'index.php' in case of no file found - it's >> simply invalid since has no leading '/'). Probably you mean to use >> >> try_files $uri /administrator/index.php; >> >> instead, and probably it's a good idea to specify separate location for >> /administrator/index.php to avoid internal redirection loop in case >> there will be no /administrator/index.php file. >> > There is no /administrator/index.php file. > I want to simply separate the administrator backend, and serve it from > the same machine as balancer, because i've got all the gallery photos on > this machine. So there is no /disk0/vhosts/domain/public/administrator directory at all, am I right? > I'm migrating this from apache. > > .htaccess says: > RewriteEngine On > RewriteCond %{REQUEST_FILENAME} -s [OR] > RewriteCond %{REQUEST_FILENAME} -l [OR] > RewriteCond %{REQUEST_FILENAME} -d > RewriteRule ^.*$ - [NC,L] > RewriteRule ^.*$ index.php [NC,L] As far as I understand you basically need to pass everything to fastcgi while rewriting everything that doesn't exist to index.php? Try something like this: root /disk0/vhosts/domain/public; fastcgi_index index.php; location / { try_files $uri $uri/ @fallback; fastcgi_pass apps; fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name; ... } location @fallback { fastcgi_pass apps; fastcgi_param SCRIPT_FILENAME $document_root/index.php; ... } location /administrator/ { rewrite ^/administrator(.*) $1 break; try_files $uri $uri/ @adminfallback; fastcgi_pass 127.0.0.1:3000; fastcgi_param SCRIPT_FILENAME $document_root$fastcgi_script_name; ... } location @adminfallback { fastcgi_pass 127.0.0.1:3000; fastcgi_param SCRIPT_FILENAME $document_root/index.php; ... } >>> fastcgi_pass 127.0.0.1:3000; >>> #fastcgi_index index.php; >>> #fastcgi_param SCRIPT_FILENAME >>> /disk0/vhosts/domain/public/index.php; >>> fastcgi_param SCRIPT_FILENAME >>> /disk0/vhosts/domain/public/$fastcgi_script_name; >>> include fastcgi_params; >>> } >>> >>> location /js/lib/xinha/plugins/ExtendedFileManager/backend.php { >>> fastcgi_pass 127.0.0.1:3000; >>> fastcgi_index backend.php; >>> fastcgi_param SCRIPT_FILENAME >>> /disk0/vhosts/domain/public/js/lib/xinha/plugins/ExtendedFileManager/backend.php; >>> include fastcgi_params; >>> access_log /var/log/nginx/backend1.log; >>> } >>> >>> location /js/lib/xinha/plugins/ImageManager/backend.php { >>> fastcgi_pass 127.0.0.1:3000; >>> fastcgi_index backend.php; >>> fastcgi_param SCRIPT_FILENAME >>> /disk0/vhosts/domain/public/js/lib/xinha/plugins/ImageManager/backend.php; >>> include fastcgi_params; >>> access_log /var/log/nginx/backend2.log; >>> } >>> >>> location ~ ^/(js)/ { >>> >> >> Any reasons to use regex here? It should be enough to use normal >> >> location /js/ { >> >> unless you want to render the above '/js/.../backend.php' locations >> useless. >> > Yes the regex is useless in this case. > Is there any way of merging this to */backend.php locations? Do not understand question. All 3 locations mentioned are quite different in your config - /js/ serves static, while backend.php's are passed to fastcgi (and one backend.php writes backend1.log, while other - backend2.log). What do you mean by "merging"? If you want to avoid duplicate parts of configuration - use include directive or inheritance from upper levels. Maxim Dounin From nginx-forum at nginx.us Thu Jul 16 23:25:52 2009 From: nginx-forum at nginx.us (gerryw) Date: Thu, 16 Jul 2009 15:25:52 -0400 Subject: FastCGI application not receiving multi-part data In-Reply-To: <6474272efa8edf4f503a18f5fe75a2cf.NginxMailingList@forum.nginx.org> References: <6474272efa8edf4f503a18f5fe75a2cf.NginxMailingList@forum.nginx.org> Message-ID: <559101f3ab5f601283c1b723388acc62.NginxMailingList@forum.nginx.org> Hello All, It looks like the config options Daniel recommended are sending the data. I think the multipart data in this case is something that most of the C/C++ cgi libs are choking on. If I just read the data from stdin and dump it, I can see the data I need. I'm going to try writing my own little parser for this. I'm using the very cool Upload module, so I don't need the actual file data. I appreciate the help y'all have given me with this. I'll come back and post the result. Thanks, Gerry Posted at Nginx Forum: http://forum.nginx.org/read.php?2,4091,4125#msg-4125 From nginx-forum at nginx.us Fri Jul 17 00:19:05 2009 From: nginx-forum at nginx.us (valve) Date: Thu, 16 Jul 2009 16:19:05 -0400 Subject: issue with XSLT [0.8.x] Message-ID: I'm having a weird issue with the XSLT module in nginx: after some random time, requesting an XSLT-transformed URL generates 500-errors. The errorlog tells me that the XML-document is not well formed, which it is. It seems to be similar to an issue raised on the Russian maillinglist (http://www.lexa.ru/nginx-ru/msg25325.html ), although I'm not using FastCGI but just use static XML files as input for the XSLT-engine. Running an strace or debugging with http_debug doesn't show anything weird as far as I can tell: 2009/07/16 21:49:34 16452#0: *65 http cl:-1 max:1048576 2009/07/16 21:49:34 16452#0: *65 generic phase: 2 2009/07/16 21:49:34 16452#0: *65 post rewrite phase: 3 2009/07/16 21:49:34 16452#0: *65 generic phase: 4 2009/07/16 21:49:34 16452#0: *65 generic phase: 5 2009/07/16 21:49:34 16452#0: *65 access phase: 6 2009/07/16 21:49:34 16452#0: *65 access phase: 7 2009/07/16 21:49:34 16452#0: *65 post access phase: 8 2009/07/16 21:49:34 16452#0: *65 content phase: 9 2009/07/16 21:49:34 16452#0: *65 open index "/var/www/allyourbass.org/htdocs/index.xml" 2009/07/16 21:49:34 16452#0: *65 internal redirect: "/index.xml?" 2009/07/16 21:49:34 16452#0: *65 http cl:-1 max:1048576 2009/07/16 21:49:34 16452#0: *65 generic phase: 2 2009/07/16 21:49:34 16452#0: *65 post rewrite phase: 3 2009/07/16 21:49:34 16452#0: *65 generic phase: 4 2009/07/16 21:49:34 16452#0: *65 generic phase: 5 2009/07/16 21:49:34 16452#0: *65 access phase: 6 2009/07/16 21:49:34 16452#0: *65 access phase: 7 2009/07/16 21:49:34 16452#0: *65 post access phase: 8 2009/07/16 21:49:34 16452#0: *65 content phase: 9 2009/07/16 21:49:34 16452#0: *65 content phase: 10 2009/07/16 21:49:34 16452#0: *65 content phase: 11 2009/07/16 21:49:34 16452#0: *65 http filename: "/var/www/allyourbass.org/htdocs/index.xml" 2009/07/16 21:49:34 16452#0: *65 http static fd: 16 2009/07/16 21:49:34 16452#0: *65 http set discard body 2009/07/16 21:49:34 16452#0: *65 xslt filter header 2009/07/16 21:49:34 16452#0: *65 http output filter "/index.xml?" 2009/07/16 21:49:34 16452#0: *65 copy filter: "/index.xml?" 2009/07/16 21:49:34 16452#0: *65 xslt filter body 2009/07/16 21:49:34 16452#0: *65 not well formed XML document while sending response to client, client: 83.161.41.147, server: klep.name, request: "GET / HTTP/1.1", host: "klep.name" 2009/07/16 21:49:34 16452#0: *65 http special response: 500, "/index.xml?" 2009/07/16 21:49:34 16452#0: *65 http set discard body 2009/07/16 21:49:34 16452#0: *65 xslt filter header 2009/07/16 21:49:34 16452#0: *65 HTTP/1.1 500 Internal Server Error Server: nginx/0.8.5 Date: Thu, 16 Jul 2009 19:49:34 GMT Content-Type: text/html Content-Length: 192 Connection: close 2009/07/16 21:49:34 16452#0: *65 http write filter: l:0 f:0 s:161 2009/07/16 21:49:34 16452#0: *65 http output filter "/index.xml?" 2009/07/16 21:49:34 16452#0: *65 copy filter: "/index.xml?" 2009/07/16 21:49:34 16452#0: *65 xslt filter body 2009/07/16 21:49:34 16452#0: *65 http postpone filter "/index.xml?" 098EDFE0 2009/07/16 21:49:34 16452#0: *65 http write filter: l:1 f:0 s:353 2009/07/16 21:49:34 16452#0: *65 http write filter limit 0 2009/07/16 21:49:34 16452#0: *65 http write filter 00000000 2009/07/16 21:49:34 16452#0: *65 copy filter: 0 "/index.xml?" 2009/07/16 21:49:34 16452#0: *65 copy filter: -1 "/index.xml?" 2009/07/16 21:49:34 16452#0: *65 http finalize request: -1, "/index.xml?" 1 2009/07/16 21:49:34 16452#0: *65 http close request 2009/07/16 21:49:34 16452#0: *65 http log handler 2009/07/16 21:49:34 16452#0: *65 close http connection: 15 Some more info $ nginx -V nginx version: nginx/0.8.5 configure arguments: --with-http_xslt_module --with-mail --http-client-body-temp-path=/tmp/nginx/client-body-temp/ --http-proxy-temp-path=/tmp/nginx/proxy-temp/ --http-fastcgi-temp-path=/tmp/nginx/fastcgi-temp/ --with-md5-asm --with-sha1-asm --with-http_ssl_module --with-http_perl_module --with-http_sub_module --with-http_secure_link_module --prefix=/usr --conf-path=/etc/nginx/nginx.conf --error-log-path=/var/log/nginx/error.log --pid-path=/var/run/nginx.pid --http-log-path=/var/log/nginx/access.log --with-debug $ uname -a Linux webserver 2.6.26-1-openvz-686 #1 SMP Sat Nov 8 21:19:32 UTC 2008 i686 GNU/Linux Does anyone have any idea what might be the issue? I've looked through the code of the module but nothing strikes me as obviously wrong. Regards, Robert Posted at Nginx Forum: http://forum.nginx.org/read.php?2,4128,4128#msg-4128 From theglobeisnowdigital at gmail.com Fri Jul 17 01:02:40 2009 From: theglobeisnowdigital at gmail.com (Magdalena Glinkowski) Date: Thu, 16 Jul 2009 14:02:40 -0700 Subject: HTTP_HEADER parsing, Documentation implies $http_range should work. Message-ID: <432c29390907161402h21ced551ree053147c9f59b1@mail.gmail.com> http://wiki.nginx.org/NginxHttpCoreModule#.24args In the documentation above for the http_HEADER section, their is implication that if I want "Range" from header, I use "http_range" as the value, is this correct? Magdalena -------------- next part -------------- An HTML attachment was scrubbed... URL: From yazz.atlas at gmail.com Fri Jul 17 03:25:16 2009 From: yazz.atlas at gmail.com (Yazz D. Atlas) Date: Thu, 16 Jul 2009 16:25:16 -0700 Subject: Help with autoindex on; missing slash for directories In-Reply-To: <20090703084935.GY5360@mdounin.ru> References: <99e8cb160906302309q37c5b209o17fd68b1f52bbdb@mail.gmail.com> <20090701093047.GC5360@mdounin.ru> <4A4B935D.3030803@gmail.com> <20090701225352.GJ5360@mdounin.ru> <4A4CF604.5030508@gmail.com> <20090702192251.GX5360@mdounin.ru> <5bf4451d0907022055y33964155m9d49df06b2a3e9ed@mail.gmail.com> <20090703084935.GY5360@mdounin.ru> Message-ID: <94C7F040-A04E-43C6-AF19-CA41F098240F@gmail.com> Maxim, Ok finally added the patch with the I believe the right stuff. Attached is the patch and the log file. I just ran "GET http://localhost:88/static" -Yazz -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: patch-nginx-debug-autoindex-ver-02.txt URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: nginx-autoindex-debug.log Type: application/octet-stream Size: 12587 bytes Desc: not available URL: -------------- next part -------------- On Jul 3, 2009, at 1:49 AM, Maxim Dounin wrote: > Hello! > > On Fri, Jul 03, 2009 at 11:55:22AM +0800, Joshua Zhu wrote: > >> Hi Maxim! >> >>> Could you please compile nginx with the attached simple patch and >>> obtain debug log for autoindex request? >>> >> >> IMHO, the ngx_log_debug1 in your patch should be changed to >> ngx_log_debug2 :) > > Yes, you are right, I missed it. But it doesn't matter as long as > nginx > compiled --with-debug. And this in turn is required since I need > this patch just for more verbose debug log on particular system. :) > > Maxim Dounin > From ek at kuramoto.org Fri Jul 17 06:14:57 2009 From: ek at kuramoto.org (Kuramoto Eiji) Date: Fri, 17 Jul 2009 11:14:57 +0900 Subject: issue with XSLT [0.8.x] In-Reply-To: