Hi, I'm new to php-fpm so I'm not sure what this notices from log are. It seems like php-fpm children dies in low traffic. It is normal ? My site have no traffic, except mine (is in testing). I configured 4 vhosts with nginx, only one has traffic (my traffic). For every vhost I configured a pool, every pool has 2 php-cgi children configured (8 totally)

[code]
Oct 01 13:56:37.121899 [NOTICE] fpm_got_signal(), line 48: received SIGCHLD
Oct 01 13:56:37.122029 [WARNING] fpm_children_bury(), line 215: child 2910 (pool default) exited on signal 11 SIGSEGV after 91618.466533 seconds from start
Oct 01 13:56:37.123379 [NOTICE] fpm_children_make(), line 352: child 3345 (pool default) started
Oct 02 18:19:39.204831 [NOTICE] fpm_got_signal(), line 48: received SIGCHLD
Oct 02 18:19:39.204921 [NOTICE] fpm_children_bury(), line 215: child 2909 (pool default) exited with code 0 after 193782.743856 seconds from start
Oct 02 18:19:39.206420 [NOTICE] fpm_children_make(), line 352: child 3686 (pool default) started
Oct 03 12:16:43.736380 [NOTICE] fpm_got_signal(), line 48: received SIGCHLD
Oct 03 19:16:43.736529 [NOTICE] fpm_children_bury(), line 215: child 3345 (pool default) exited with code 0 after 166777.387967 seconds from start
Oct 03 19:16:43.737681 [NOTICE] fpm_children_make(), line 352: child 4006 (pool default) started
Oct 04 17:06:16.498091 [NOTICE] fpm_got_signal(), line 48: received SIGCHLD
Oct 04 17:06:16.498212 [NOTICE] fpm_children_bury(), line 215: child 3686 (pool default) exited with code 0 after 143172.157992 seconds from start
Oct 04 17:06:16.499810 [NOTICE] fpm_children_make(), line 352: child 4428 (pool default) started

[/code]

Thanks.

On Sun, Oct 4, 2009 at 6:28 PM, ktm <nginx-forum@nginx.us> wrote:
>
> Hi, I'm new to php-fpm so I'm not sure what this notices from log are. It seems like php-fpm children dies in low traffic. It is normal

There's a setting which controls this. (Default = 500). So after
serving n=500 requests, the child worker process will voluntarily exit
itself. The manager above should re-fork a new child to replace the
exited one. This is normal behaviour. The child will kill itself after
n requests because to stop memory leaks building up.

dreamcat4
dreamcat4@gmail.com