[ClusterLabs] DRBD resource not failing over upon putting node into standby

[Simon Lawrence]

On 26/02/15 14:30, Simon Lawrence wrote:
> Hi
>
> I am in the early stages of building a 2 node cluster running 
> Centos6.6/DRBD8.4/Cman/Pacemaker1.1.12
>
> So far I have just two resources: a drbd volume (centos1_drbd) and a 
> KVM VM (centos1) that uses the drbd volume.
> The /shared filesystem containing the VM config is currently manually 
> replicated on both nodes.
>
> The cluster correctly brings up the drbd resouce in primary/secondary 
> mode. It is fully synced.
>
> I have colocation and order constraints relating to the DRBD and VM 
> resources, so the VM starts up correctly on the master node.
>
> I am able to move VM from node to node using 'pcs resource move 
> centos1 <node>'. The drbd resource gets promoted/demoted accordingly.
>
> However if I put one of the nodes into standby ('pcs cluster standby 
> node1.bcn'), the drbd resource does not failover.
>


I have stumbled across the cause of the issue.

I have multiple fqdn's in my hosts files for my nodes to differentiate 
between the various network interfaces.
The nodes names in my drbd config didn't exactly match the node names 
I'd used in the cluster config.

May be obvious to some, but I've never seen it explicitly stated that 
they should match (and I've read a lot of documentation and howto's)