Hello Maxim, until now no signal 11 occured without substitutions4nginx. It's interesting that such errors do not occur in 1.0. I'll contact the developer of this module, maybe he can help or fix it. We're parsing the content with regexp to rewrite links to another domain, mainly for external CDN domains. Thank you very much for your help.by mscheel - Nginx Mailing List - English
Thanks. Do you think it's necessary to create a new rpm without this module or will it be sufficient to turn it off in the configs? I'll disable the config now.by mscheel - Nginx Mailing List - English
Interesting, i'm unable to reproduce it with the uri. The uri generates a pdf file (when it does not exist on the fs). I've deleted it on the fs and it's created fine by our php application. I've also made some tests with keepalive off, no change. Maybe there is something missing in SLES 10 which is required by 1.1? A manual configure and make (without install) with my nginx -V runs without errorby mscheel - Nginx Mailing List - English
Hello Maxim, thanks alot for your help. I'll try to reproduce it. (gdb) fr 0 #0 0x000000000042cbe8 in ngx_readv_chain (c=0x2b508a1d43d0, chain=0x602348) at src/os/unix/ngx_readv_chain.c:195 195 src/os/unix/ngx_readv_chain.c: No such file or directory. in src/os/unix/ngx_readv_chain.c (gdb) p *chain->buf $6 = {pos = 0x0, last = 0x0, file_pos = 0, file_last = 0, start = 0by mscheel - Nginx Mailing List - English
Hello Maxim, it appears only on two productive servers. Compiling without these modules is not possible at the moment but i've tried several times with each module deactivated (and without any of these three). Full config is huge. I'm using a php-fpm backend with keepalive on and one upstream server with nginx and php-fpm. http { include mime.types; include gzip.conf; default_typeby mscheel - Nginx Mailing List - English
sorry, forgot uname -a: Linux hostname 2.6.16.60-0.42.10-smp #1 SMP Tue Apr 27 05:11:27 UTC 2010 x86_64 x86_64 x86_64 GNU/Linux nginx -V nginx version: nginx/1.1.19 built by gcc 4.1.2 20070115 (SUSE Linux) TLS SNI support disabled configure arguments: --add-module=/root/rpm/SOURCES/agentzh-headers-more-nginx-module-3580526 --add-module=/root/rpm/SOURCES/FRiCKLE-ngx_cache_purge-e34bd2c -by mscheel - Nginx Mailing List - English
Dear all, error logs show this error (worker process exited on signal 11). The error appears after ~1 hour, then with the half of the worker processes. No problems with 1.0-versions of nginx. Problem existed in 1.1.18 already. I'm using a custom rpm build with the following additional modules: --add-module=/root/rpm/SOURCES/agentzh-headers-more-nginx-module-3580526 \ --add-module=/root/rpmby mscheel - Nginx Mailing List - English
Your right, setting it for fastcgi as param works fine. I had a problem in the past setting variables in the fastcgi_params, they've just been ignored. This seems to be gone. Then i have to rewrite the nginx internal logic of my substitions/mappings based on this variable. I still think allowing a manual override of this variable should be allowed. If any other plugin relies on this header useby mscheel - Nginx Mailing List - English
Antonio P.P. Almeida Wrote: ------------------------------------------------------- > > Dear all, > > > > starting with version 1.1.11 the $https variable > is activated by > > default. This is good as many applications rely > on them because apache > > is using this header. > > However, now i cannot set or map this variable > (nginx: the &gby mscheel - Nginx Mailing List - English
Dear all, starting with version 1.1.11 the $https variable is activated by default. This is good as many applications rely on them because apache is using this header. However, now i cannot set or map this variable (nginx: the duplicate "https" variable). We have a load balancer in front of the web servers which does all SSL traffic. The nginx servers can only differ between http anby mscheel - Nginx Mailing List - English