[Pacemaker] Not unmoving colocated resources can provoke DRBD split-brain

[Digimer]

On 22/05/14 12:08 PM, Robert Dahlem wrote:
> Hi,
>
> On 22.05.2014 16:59, Digimer wrote:
>
>> I can't speak to the pacemaker issue, but I can say that a proper
>> stonith config in pacemaker and fencing config in drbd would prevent a
>> split-brain. This would cause a node to reboot in this scenario, so you
>> still need to resolve it, but a reboot is a heck of a lot better than a
>> split-brain.
>
> And what exactly would be the sufficient cause for fencing in this scenario?
>
> Regards,
> Robert

If DRBD detects a split-brain is about to happen, it will block and call 
it's fence handler (fencing resource-and-stonith;). In pacemaker setups, 
this is done via the crm-fence-peer.sh handler. It will take the DRBD 
fence request and pass it up to pacemaker. Once pacemaker confirms the 
peer is fenced, the surviving node will unblock and resume operation, 
thus avoiding a split-brain.

This still leaves the cause of the initial issue unresolved, but it does 
make the consequences of that issue far less damaging.

-- 
Digimer
Papers and Projects: https://alteeve.ca/w/
What if the cure for cancer is trapped in the mind of a person without 
access to education?