[Pacemaker] Pacemaker resource migration behaviour

[Andrew Beekhof]

On Thu, Mar 7, 2013 at 11:23 AM, Andrew Beekhof <andrew at beekhof.net> wrote:
> On Wed, Mar 6, 2013 at 8:02 PM, James Guthrie <jag at open.ch> wrote:
>> Hi Andrew,
>>
>> Thanks for looking into this. We have since decided not to perform a failover on the failure of one of the sub-* resources for operational reasons. As a result, I can't reliably test if this issue is actually fixed in the current HEAD. (Speaking of which, do you have a date set yet for 1.1.9?)

ASAP.  I'm hoping in a couple of hours from now, otherwise tomorrow.

>>
>> On Mar 6, 2013, at 8:39 AM, Andrew Beekhof <andrew at beekhof.net> wrote:
>>
>>> I'm still very confused about why you're using master/slave though.
>>
>> The reason I went with master-slave was that we want the init script started on the "master" host and stopped on the "slave".
>
> You get those semantics from a normal primitive resource.
>
>> With a master-slave I have a monitor operation on the slave ensuring that the resource will be stopped on the slave if it were to be started manually (something I can't be sure wouldn't happen). AFAIK this wouldn't be the case with a "standard" resource.
>
> I think 1.1.8 allowed for operations with role=Stopped which would do
> this for the highly paranoid :-)
>
>>
>> Regards,
>> James
>>
>>
>> _______________________________________________
>> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
>> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>>
>> Project Home: http://www.clusterlabs.org
>> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
>> Bugs: http://bugs.clusterlabs.org