[ClusterLabs] Stopping all nodes causes servers to migrate

[Ken Gaillot]

On Mon, 2021-01-25 at 14:01 +0300, Andrei Borzenkov wrote:
> On Mon, Jan 25, 2021 at 12:07 PM Jehan-Guillaume de Rorthais
> <jgdr at dalibo.com> wrote:
> 
> > As actions during a cluster shutdown cannot be handled in the same
> > transition
> > for each nodes, I usually add a step to disable all resources using
> > property
> > "stop-all-resources" before shutting down the cluster:
> > 
> >   pcs property set stop-all-resources=true
> >   pcs cluster stop --all
> > 
> 
> How exactly does it interact with target-role of each individual
> resource? What has precedence? Does this option stop resources once
> or
> it will prevent activation of resources when the cluster is
> restarted?
> 
> Sorry for questions, but there is no documentation really beyond that
> "stop-all-resources" will stop all resources.

We do need to add that to the documentation ...

In order of highest precedence first:

  is-managed=false (Pacemaker won't stop or start)
  stop-all-resources=true
  target-role on a specific resource
  target-role in resource defaults

stop-all-resources will keep resources stopped for as long as it's set
to true. It has to be set back to false (or removed) for resources to
start again.
-- 
Ken Gaillot <kgaillot at redhat.com>