[Pacemaker] resource with colocation rule doesn't fail

[Johan Huysmans]

On 02-08-13 03:33, Andrew Beekhof wrote:
> On 01/08/2013, at 5:38 PM, Johan Huysmans <johan.huysmans at inuits.be> wrote:
>
>> I forgot to mention:
>>
>> I'm using a build from git (Version: 1.1.11-1.el6-42f2063).
>> I used the same config on an old 1.1.10 rc (rc6 or before) and that worked, as of rc7 it didn't work anymore.
> I will have a look, but why are you setting on-fail=block for everything?
I'm using the on-fail=block for my cloned resources as I want to achieve 
following setup:

I always have 2 node setups, on these nodes some services have to run on 
both nodes, some services only have to run on 1 node.
The services running on 1 node must run on the node where the other 
services are running correctly.
The services that need to be running on both nodes are configured as 
cloned resources, the services only running on 1 node are grouped in a 
resource group. An colocation rule configures that the resource group 
must run where the cloned resource is running.

We configure the cloned resources as on-fail=block as we don't want 
these services to be restarted when something happens.
Once the resource recovers (automatically) the cluster should pick this 
up and allow the resource group to run on that node again.
This way we don't have to perform any pacemaker related actions every 
time a problem occurs.

I said "cluster _should_ pick this up", as there were some issues with 
this use-case, however the many improvements fixed a lot a issue and we 
are almost there ;)


Gr.
Johan

>
>> On 01-08-13 09:35, Johan Huysmans wrote:
>>> Hi,
>>>
>>> I have a cloned resource and a resource group. They have a colocation rule configured in such way that the resource group must be running where the cloned resource is running.
>>>
>>> When I stop the service in the cloned resource on 1 node, this appears in crm_mon, however the resource group does not get moved to the other node.
>>>
>>> When I start the service in the cloned resource on that specific node, this gets updated in crm_mon, however this still appears in the logs as stopped.
>>>
>>> I attached the crm_report for this problem.
>>>
>>> Thx!
>>>
>>> gr.
>>> Johan
>>
>> _______________________________________________
>> 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
>
> _______________________________________________
> 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