[Pacemaker] remaining node hang & sysconfig/cman CMAN_LEAVE_OPTS="remove"

[Andrew Beekhof]

On Tue, Aug 21, 2012 at 6:07 AM, Bob Haxo <bhaxo at sgi.com> wrote:
> Greetings,
>
> Is there a reason that the setting the sysconfig/cman variable
> CMAN_LEAVE_OPTS="remove" is not documented or even mentioned other than
> in man cman_tool?  A Google search for "CMAN_LEAVE_OPTS Pacemaker"
> generates a serious paucity of hits (4), and there have been 0 instances
> of the variable name in the past several years on this list.

On this list cman is hardly ever mentioned.  There is some cross-over
between lists, but for the most part, the cman experts are on
linux-cluster at redhat.com
Personally I've never seen this option before in my life :)

>
> >From man cman_tool:
>
> remove
>     Tells the rest of the cluster to recalculate quorum such that
>     activity can continue without this node.
>
> Setting this variable to "remove" seems to fix the problem of the
> surviving node cluster functions hanging when the other node is halted
> or rebooted.
>
> For the record, "service cman stop remove" works.
>
> The Pacemaker no-quorum-policy="ignore" seems to not affect this
> behavior.  One could imagine that setting that policy on a
> CMAN/Pacemaker cluster would also set CMAN_LEAVE_OPTS="remove".

No, no-quorum-policy only affects the Pacemaker layer - instructing it
to ignore whatever quorum result cman/corosync/heartbeat/etc comes up
with.

>
> Cheers,
> Bob Haxo
>
>
> _______________________________________________
> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://bugs.clusterlabs.org