I have a 3 node cluster with CRM.  One node is configure for quorum
only.  I have the Default Resource Stickiness set to INFINITY for the
cluster.  When I take the quorum node down, the two other nodes seem
to renegotiate.  This seems completely unnecessary as the two nodes
have quorum between the two and the resource should not change.
During this process of figuring out who should keep running the
services, it tries to restart an already running named process:

Apr 24 18:54:30 dhcp1-admin lrmd: [2901]: info: RA output:
(named:start:stdout) Starting named:
Apr 24 18:54:30 dhcp1-admin lrmd: [2901]: info: RA output:
(named:start:stdout) named: already running[FAILED]
Apr 24 18:54:30 dhcp1-admin lrmd: [2901]: info: RA output:
(named:start:stdout)

Of the 8 other resources managed by heartbeat, named is the only one
it tried to restart.  This puts the entire cluster into a half working
state.  Cleaning up the resource gets the cluster back to full health.
 Why is this only happening with the named process?  Thanks for any
insights.




-- 
Jiann-Ming Su
"I have to decide between two equally frightening options.
 If I wanted to do that, I'd vote." --Duckman
"The system's broke, Hank.  The election baby has peed in
the bath water.  You got to throw 'em both out."  --Dale Gribble
"Those who vote decide nothing.
Those who count the votes decide everything.”  --Joseph Stalin
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Reply via email to