[Linux-HA] Resource won't migrate with crm_resource...

[Zachár Balázs]

I had the same problem!

I restart the heartbeat on the "failed" node, and the resources migrated
back there /In my case: (stickiness=0)=(auto_failback on)/



Brian O'Neill írta:
> I should also say that failcount on node0 for admin0_group is 0.
>
>
> Brian O'Neill wrote:
>> I'm trying to migrate a resource from its current node to it's
>> "normal" node, and crm_resource is having no effect on it.
>>
>> I have a three-node cluster running 2.0.7. node0 is the normal node
>> for this service (with a locational constraint of 200), node2 is the
>> secondary node (constraint of 100), and node1 is the node of last
>> resort (constraint of 0). The current cib.xml shows no other
>> constraints in effect. symmetric_cluster is true, and
>> default_resource_stickiness is INFINITY (which may be the problem - I
>> may not understand the implication there).
>>
>> The resource is called admin0_group, and consists of an sshd instance
>> and an apache instance.
>>
>> I'm executing "crm_resource -M -r admin0_group -H node0" on node0.
>> I've used this to move the resource around before.
>>
>> I see it set a location constraint on node0 of INFINITY for
>> admin0_group, but never move the resource.
>>
>> In the log on node1 (the current active node), it only mentions the
>> setting of the constraint. Nothing is logged regarding attempting to
>> shut down the resources.
>>
>> On node0, it appears to fail with this:
>>
>> Sep 18 09:11:58 node0 pengine: [6930]: WARN: unpack_rsc_op:unpack.c
>> Processing failed op (admin0_chroot_sshd_stop_0) for
>> admin0_chroot_sshd on node0
>> Sep 18 09:11:58 node0 pengine: [6930]: WARN: unpack_rsc_op:unpack.c
>> Handling failed stop for admin0_chroot_sshd on node0
>>
>> I checked - running the sshd OCF script on node0 (where the sshd
>> instance is not currently running) exits with a status of 0, so I
>> don't know why the crm is considering it failed.
>>
>> No attempt to shut it down on node1 seems to be made - I assume it is
>> checking to make sure it is not already running on node0.
>>
>> Any ideas?
>>
>> -Brian
>> _______________________________________________
>> Linux-HA mailing list
>> Linux-HA at lists.linux-ha.org
>> http://lists.linux-ha.org/mailman/listinfo/linux-ha
>> See also: http://linux-ha.org/ReportingProblems
> _______________________________________________
> Linux-HA mailing list
> Linux-HA at lists.linux-ha.org
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
>
>