I will try to get more details on this issue. It is very easily
reproducible. The default corosync init file (/etc/init.d/corosync) has
"chkconfig - 20 20". The rsyslog started at 26. So I changed the starting
index of corosync to 99. Sometimes if I forget to change this starting
index, then this problem happens. I think the default starting index 20 of
corosync is too early.

Thanks
hj


On Tue, Feb 9, 2010 at 8:48 AM, Jan Friesse <[email protected]> wrote:

> Hi,
> I don't personally believe, that problem is caused by syslog turned
> on/off. It looks like some another race combined with deadlock, this
> time caused by conjunction of schedwrk serialize lock and logsys
> serialize lock.
>
> Can you please try attached patch (apply to current trunk).
>
> Regards,
>  Honza
>
> hj lee wrote:
> > I noticed that this happens when corosync starts before syslog in init
> start
> > order. I understand that corosync requires syslog, but at least it should
> > start OK and should be operational OK even without syslog.
> >
> > On Mon, Feb 8, 2010 at 10:59 AM, hj lee <[email protected]> wrote:
> >
> >> Hi,
> >>
> >> When corosync starts to fail with some reasons, multiple corosync
> processes
> >> are created. But this does not help at all. All these processes are
> stuck
> >> also. How or who does start corosync multiple time? If it does, then it
> >> should kill the corosync process before starting new corosync.
> >>
> >> Thanks
> >> hj
>
>
> diff --git a/trunk/exec/main.c b/trunk/exec/main.c
> index beedc2c..a9aa094 100644
> --- a/trunk/exec/main.c
> +++ b/trunk/exec/main.c
> @@ -156,9 +156,6 @@ static void unlink_all_completed (void)
>  {
>        poll_stop (corosync_poll_handle);
>        coroipcs_ipc_exit ();
> -       totempg_finalize ();
> -
> -       corosync_exit_error (AIS_DONE_EXIT);
>  }
>
>  void corosync_shutdown_request (void)
> @@ -1538,6 +1535,8 @@ int main (int argc, char **argv)
>         */
>        poll_run (corosync_poll_handle);
>
> +       totempg_finalize ();
> +
>        return EXIT_SUCCESS;
>  }
>
>
>


-- 
Peakpoint Service

Cluster Setup, Troubleshooting & Development
[email protected]
(303) 997-2823
_______________________________________________
Openais mailing list
[email protected]
https://lists.linux-foundation.org/mailman/listinfo/openais

Reply via email to