On 11/06/2015 03:44 AM, Karthikeyan Ramasamy wrote: > Hello Users, > We have a cluster setup with Pacemaker and are trying node addition > dynamically in maintenance mode. When coming out of maintenance mode, there > is a restart of all the services managed by Pacemaker. We see a stop and > start command issued for each of the service. This is observed in dl360 > environment. > > When we tried originally in a different environment(a sb6270 machine), > coming out of maintenance mode just sent a start command and came back with > already running status. > > Is this behavior of restarting services environment specific or are we > missing something in the new environment? (We are still checking for the > difference in steps performed). In the meantime, if some other user could > provide us hints, it would be very helpful. > > Thanks, > Karthik.
Do I understand right: you add a new node to the cluster when in maintenance mode, then when you turn maintenance mode off, all services are restarted on all nodes? I wouldn't expect that, but I've never tried it myself. Are all your resources clones that are ordered after each other, without interleave=true? Then that behavior would be expected, because to start each clone on the new node, I think it would have to stop everything that depends on it first. Or, if any services are already running on the new node, that could potentially lead Pacemaker to consider them failed, since it hasn't assigned them to it yet. That could trigger a recovery across all nodes. _______________________________________________ Users mailing list: [email protected] http://clusterlabs.org/mailman/listinfo/users Project Home: http://www.clusterlabs.org Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf Bugs: http://bugs.clusterlabs.org
