Re: apache reload and child fault

[Tomas Hoger <thoger@pobox.sk>]

On Tue, Dec 02, 2003 at 12:27:33PM +0100, Roberto Sebastiano wrote:
> /etc/init.d/apache stop; killall apache; sleep 10; /etc/init.d/apache
> start seems the right approach to workaround the problem

I have to warn you, that this might not help.  After reading Rodi's post
from september, I have written simple watchdog script, which tries to
detect apache crashes.  It looks into /var/run/apache.pid and if apache
process with given pid is not running, it collects some system info
(process list, disk free space, network connections, could this be of any
use Fabio?) and at the end does exactly the same to restart apache.  It
saved me from troubles at least one since then, but it failed today.
killall apache haven't killed all apache child processes and start failed
with 'could not bind 80'.  Actually, my script is run from cron, so it
tried again and again and it finally worked out fine at the 8th or 9th try.
Hope killall -9 will work better in the future.

-th.