In our 2-node-cluster with pacemaker 1.1.2 and corosync 1.2.1 we have defined a
resource 'auvm1' with 'on-fail=restart' for op's start and monitor and we have
set 'on-fail=restart' in op_defaults.
The resource 'auvm1' was started on node 'goat1' when we provoked an error
preventing the resource to run on that node furthermore. Due to the
on-fail-settings we expected a restart of the resource 'auvm1' on the second
node 'sheep1'. However we saw that 'auvm1' was stopped on node 'goat1' and not
restarted.
It seems to us that pacemaker ignores the operation specific on-fail-settings
for the resource 'auvm1' and considers only the op_defaults (for stop op :
restart, for start op and monitor op : stop ?).
1) How can we achieve that the operation-specific-settings for the resource
have a higher priority than
the default-setting ? We'd like to have the resource restarted on the
second node without fencing
the first node.
2) Is it possible to define different op_defaults for op stop, op start and op
monitor ?
Regards,
Armin Haussecker
Extract from our CIB:
crm configure show
node goat1
node sheep1
primitive auvm1 ocf:heartbeat:Xen \
op start on-fail="restart" start-delay="0s" \
op monitor on-fail="restart" interval="60s" \
op stop on-fail="block" \
meta is-managed="true" resource-stickiness="1000"
migration-threshold="2" priority="300"
target-role="Stopped" \
params xmfile="/home/clusteradm/cluster/xen/auvm1"
..............
..............
op_defaults $id="op-options" \
on-fail="restart"
_______________________________________________
Openais mailing list
[email protected]
https://lists.linux-foundation.org/mailman/listinfo/openais
