Re: FastCGI process dies without being noticed and respawned

We have been running a system since 2009.. not sure I have ever seen this m=
essage but for my two cents worth=85

Is it possible that a fastcgi process is getting sent a message from manage=
r when it is in the middle of processing another; the manager thinks it is =
unresponsive, and then kills it?

Or=85 could the file handles still be open/with-data from a previous reques=
t that was not flushed?

Would need to look at the access/error logs from before and after I suspect.

Tom

On Jul 3, 2013, at 3:40 AM, mephinet <[email protected]> wrote:

> Dear FastCGI devs,
> =

> we have a fastcgi program (written in C++) in production for several year=
s now. On one of our systems, we see this phenomenon, that the process sudd=
enly dies without any information in the logs and without being respawned.
> =

> Normally, respawning works fine:
> =

> [Tue Jul 02 07:55:19 2013] [warn] FastCGI: server ".../catcgi.fcgi" resta=
rted (pid 9274)
> [Tue Jul 02 08:08:57 2013] [warn] FastCGI: server ".../catcgi.fcgi" resta=
rted (pid 8632)
> [Tue Jul 02 15:01:53 2013] [warn] FastCGI: server ".../catcgi.fcgi" resta=
rted (pid 28313)
> [Tue Jul 02 15:04:02 2013] [warn] FastCGI: server ".../catcgi.fcgi" resta=
rted (pid 1906)
> =

> however, two hours later:
> =

> [Tue Jul 02 16:53:11 2013] [error] [client 194.158.159.8] (111)Connection=
 refused: FastCGI: failed to connect to server ".../catcgi.fcgi": connect()=
 failed
> [Tue Jul 02 16:53:11 2013] [error] [client 194.158.159.8] FastCGI: incomp=
lete headers (0 bytes) received from server ".../catcgi.fcgi"
> [Tue Jul 02 16:53:11 2013] [error] [client 194.158.159.8] (111)Connection=
 refused: FastCGI: failed to connect to server ".../catcgi.fcgi": connect()=
 failed
> [Tue Jul 02 16:53:11 2013] [error] [client 194.158.159.8] FastCGI: incomp=
lete headers (0 bytes) received from server ".../catcgi.fcgi"
> ...
> =

> The server configuration:
> =

> LogLevel                warn
> FastCgiServer .../catcgi.fcgi \
>     -initial-env ... \
>     -idle-timeout 120
> =

> How can that happen that the fastcgi process dies (or ceases to accept co=
nnects) without the fastcgi process manager taking notice, logging an error=
 and respawning it?
> =

> The system is:
> Red Hat Enterprise Linux Server release 5.6 (Tikanga)
> Apache/2.2.3
> mod_fastcgi/2.4.6
> Linux 2.6.18-238.1.1.el5
> =

> Thanks for your support!
> Mephinet
> _______________________________________________
> FastCGI-developers mailing list
> FastCGI-developers-xGejAJT2w6xVgU18Zptdi0EOCMrvLtNR@public.gmane.org
> http://mailman.fastcgi.com/mailman/listinfo/fastcgi-developers