[Pacemaker] cleaning resources restart dependents

Andreas Kurz andreas.kurz at linbit.com
Fri Aug 6 12:00:21 UTC 2010


On 2010-08-06 13:49, Schaefer, Diane E wrote:
>  >> Hi,
>
>  >>
>
>  >> ? We had a question as to why the dependent resources are restarted
> when a
>
>  >> clean is issued to the parent to clear the fail-count if all were
> happily
>
>  >> running?? This is with pacemaker 1.0.6 and heartbeat 3.0.
>
>  >Clean removes the resource's operation history.
>
>  >The cluster then asks all nodes whether they have the resource running.
>
>  From the trace it looks like the parent is sent a monitor:0 request,
> while each child is sent a stop.
>
> It doesn’t seem to wait until the monitor:0 of the parent reports
> running before sending out the stop, so each dependent child will stop
> first before restarting.

Looks like an already fixed bug:

http://hg.clusterlabs.org/pacemaker/stable-1.0/rev/3b95166232d4

Regards,
Andreas




More information about the Pacemaker mailing list