PHP-FPM seg faults, how to find out why

PHPphp-fpmsegmentation-fault

I have a Nginx running php through PHP-FPM. When I do a phpinfo() the site shows up just fine but when i run my php code (which works fine on a VM that I have, this is an amazon EC2 instance) it seg faults. The error message is not very descriptive. Is there a way to find out WHY my code is causing php-fpm to seg fault?

This is the error from the php-fpm error.log

[12-May-2011 21:04:45] WARNING: [pool www] child 2447 exited on signal 11 (SIGSEGV) after 4026.581320 seconds from start
[12-May-2011 21:04:45] NOTICE: [pool www] child 2623 started

Best Answer

You could try to find the line of code where php quits using XDebug.

Related Solutions

Nginx – Ever increasing mem usage by php-fpm

Some ideas:

Lower your fpm settings then increase as needed:

pm.max_children = 10 # concurrent children's
pm.start_servers = 2
pm.min_spare_servers = 2
pm.max_spare_servers = 4
pm.max_requests = 300 # till restart

Lower the nginx settigns.
Install pecl-apc(Debian package is php5-apc).
Update your php/nginx version to latest stable

What is your server spec?

PHP crashing (seg-fault) under mod_fcgi, apache

Getting a core dump involves setting the "core file size" rlimit to something non-zero for the process(es) you want to get core dumps from. Given that you're running PHP in FCGI, you don't really care about Apache itself, it's not segfaulting; you want to have a script to run ulimit -c unlimited then exec your FCGI server.

I like to have a separate scratch partition for cores, so if they get a bit big or unruly they don't cause disk fill problems elsewhere -- set /proc/sys/kernel/core_pattern to something like /var/cores/%e-%p.core.

Best Answer

Related Solutions

Nginx – Ever increasing mem usage by php-fpm

PHP crashing (seg-fault) under mod_fcgi, apache

Related Topic