[Linux-HA] About the problem that failcount is ignored
beekhof at gmail.com
Fri Sep 21 01:13:52 MDT 2007
On 9/21/07, HIDEO YAMAUCHI <renayama19661014 at ybb.ne.jp> wrote:
> I constituted a cluster in two nodes.
> I operated the next...
> 1)I cause a monitor error in an active node(rh44-1).
> 2)failcount of the active node becomes 1.
> 3)A resource moves to a standby node(rh44-2).
> 4)I stop a standby node(rh44-2).
> 5)The resource does not run anywhere.
> 6)I pushed the CLENAUP RESOURCE button with GUI(hb_gui).
> 7)failcount is 1, but a resource runs in the first
> I think that it is strange that the resource starts in the
> first node till failcount clears it.
> How do you think?
I noticed this the other day when testing the no colocation code.
The problem is that the fail-count for rscX on nodeY isn't applied
when there are no actions for rscX on nodeY (which happens when you
click the cleanup button).
If you'd like to log a bug for it, I'll make sure it gets fixed.
More information about the Linux-HA