[ClusterLabs] Coming in Pacemaker 1.1.15: graceful Pacemaker Remote node stops

[Ken Gaillot]

Pacemaker's upstream master branch has a new feature that will be part
of the eventual 1.1.15 release.

Since version 1.1.10, Pacemaker has supported Pacemaker Remote nodes to
scale a cluster beyond the limits of a full cluster stack (and to manage
resources inside a VM that is itself managed by the cluster). Pacemaker
Remote nodes can run resources, but do not participate in quorum or
other activities of the underlying messaging layer.

In current released versions of Pacemaker, if the pacemaker_remoted
daemon is stopped on an active Pacemaker Remote node, the cluster will
fence the node. Because updating the software requires a restart of the
service, this means that resources creating Pacemaker Remote nodes must
be stopped before the node can be updated.

With the new feature, pacemaker_remoted will notify the cluster when it
needs to stop, and the cluster will gracefully move all resources off
the node, then wait for the node to come back up. This simplifies
package updates on Pacemaker Remote nodes, and allows the process to be
scripted more easily.

This new feature makes updates of Pacemaker Remote nodes more similar to
that of cluster nodes -- simply stop cluster services (in this case
pacemaker_remoted) before updating.
-- 
Ken Gaillot <kgaillot at redhat.com>