[Pacemaker] Two nodes split brain case

Priyanka Ranjan priyanka3rdfeb at gmail.com
Wed Feb 18 08:23:33 EST 2009


Hi All,
We all know that when we get split brain situation  in two nodes (openais+
pacemaker) cluster , both nodes try  to stonith each other.  if  stonith
action is "reset" and  suppose node1 succeeds to reset node2 first . when
node2 comes up and still finds the split brain situation, it stonith node1.
and this (resetting each other) continues till the split brain is there.
to prevent this we set stonith action to poweroff or dont enable openais
service at startup.

If we compare the above openais+pacemaker cluster behaviour to RHCS , they
are handling this situation pretty well ,   there when node2 comes up again
and finds that split brain situation is still there , instead of stonithing
node1 it resets itself.  so the resetting each other in cycle does not
happens.

Does openais+pacemaker is also planning to do something like RHCS to get rid
of this problem in future. Please share your view.

Thanks and Regards,
Priyanka.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.clusterlabs.org/pipermail/pacemaker/attachments/20090218/62d4118c/attachment.html>


More information about the Pacemaker mailing list