[Pacemaker] DRBD cluster and UPS

Vladimir Elisseev vovan at vovan.nl
Thu Oct 11 01:06:40 EDT 2012


I'm trying to find a good solution to properly shutdown a DRBD
(master/slave) cluster in case of outages. The problem, at least for me,
is that if you simply shutdown one node, when it comes back in 99% on
the cases you'll end up with split brain scenarios. So for me the proper
logic in case of power failure looks like (to simplify things I call
nodes node1 and node2):
1. power failure occurs
2. wait, let say 60 sec, and then put node2 in standby mode
3. in order to minimize power consumption power off node2
4. when only 5 min of battery is left power off node1

The question is, what would be the proper way to perform the step 3
after the step 2 (you have to be sure that node 2 in standby mode before
power this node down). There might be another solutions, so I'd
appreciate any thoughts regarding this subject.


More information about the Pacemaker mailing list