2.1.2 didn't use fail-count at all for failed stop event.
they always resulted in the resource being barred from the node.
2.1.3 is well known to be broken and always set fail-count to 1
pacemaker-0.6 increments fail-count by 1 or INFINITY depending on the
value of start-failure-fatal.
but if there are any actions left over from a 2.1.2 cluster, then it
needs to enforce the 2.1.2 semantics.
hence the need for this compatibility code.
On Sep 29, 2008, at 4:11 AM, Junko IKEDA wrote:
Hi,
its regarding compatibility with the old way of handling start
actions
- ie. that failed start actions were unconditionally fatal (ie.
forced
the resource away from that machine)
Does old way mean, handling start actions error as fail-count=0?
When I used Heartbeat 2.1.2, it handled the start error as fail-
count=0.
Heartbeat 2.1.3 can handle its fail-count as 1.
Thanks,
Junko
_______________________________________________
Pacemaker mailing list
[email protected]
http://list.clusterlabs.org/mailman/listinfo/pacemaker
_______________________________________________
Pacemaker mailing list
[email protected]
http://list.clusterlabs.org/mailman/listinfo/pacemaker
