[hobbit] paging with REPEAT problem...

[Henrik Stoerner]

On Sun, Mar 27, 2005 at 12:31:12PM +0200, olivier at qalpit.com wrote:
> 
> in page.log, i see this :
> 2005-03-27 11:09:08 Worker process died with exit code 0, terminating
> 2005-03-27 11:09:08 Could not get shm of size 102400: No such file or directory
> 2005-03-27 11:09:08 Channel not available
> 2005-03-27 11:18:54 Worker process died with exit code 0, terminating
> 2005-03-27 11:18:54 Could not get shm of size 102400: No such file or directory
> 2005-03-27 11:18:54 Channel not available
> 2005-03-27 11:48:39 Worker process died with exit code 0, terminating
> 2005-03-27 12:01:01 Worker process died with exit code 0,
> terminating

Your hobbitd_alert proces dies for some reason, and when restarting it
has forgotten about when is the next time to send out an alert.

So why does it die ... the only reason I can come up with is that it
catches a signal from a child-process. Could you try changing line 332
of hobbitd/hobbitd_alert.c from
   sigaction(SIGPIPE, &sa, NULL);
to
   signal(SIGPIPE, SIG_IGN);

and let me know if that makes it keep on running ? If it does, then
the mail program that is launched to send the alerts does something
weird with it's I/O.


Henrik