Greetings,
I have a cluster with one resource group, containing two resources. If
the second resource in the group can't start (the start op fails), I end
up with one resource started, and the other stopped, i.e. we have the
following sequence of events:
1) res-1 start (succeeds)
2) res-2 start (fails)
3) res-2 stop (succeeds)
4) res-1 stop (succeeds)
5) res-1 start (succeeds)
6) res-2 start (fails)
7) ....and now we're stuck with res-1 started and res-2 stopped.
No further starts/stops occur.
Given that resource groups behave otherwise like a set of co-located
resources, I'd have expected that if res-2 continually failed to start,
that res-1 would also be forced to stop.
I tried a similar test with two separate resources, with ordering &
co-location constraints in place, and got the behaviour I expected:
1) res-1 start (succeeds)
2) res-2 start (fails)
3) res-2 stop (succeeds)
4) res-1 stop (succeeds)
5) res-1 start (succeeds)
6) res-2 start (fails)
7) res-2 stop (succeeds)
8) res-1 stop (succeeds)
9) ....and now everything is stopped, because res-2 can't run.
I'm using heartbeat 2.1.3. Can anyone shed any light on likely reasons
for this discrepancy in behaviour?
Thanks,
Tim
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
