even if all three nodes are trying to reset the errant  node, dont you think
it would be better if a node would have informed other 2 members after
fencing the errant node that i have successfully stonith the errant node. it
will save other 2 nodes in putting extra effort in  fencing the errant node
again.

i think it's even better  to elect one node among three nodes  for
stonithing the errant node and if it fails to stonith then second node
should be given a chance.  do you see any problem with this approach .

one thing more related to stonith design. suppose we have 4 nodes cluster so
do we need 3 stonith on each node ( to make a node eligible to kill any of
other 3 nodes ) that will lead to 3x4 stoniths or  do we need only  one
stonith on  each node that will be total 4 stoniths.

Thanks Andrew for your help..

On Fri, Mar 13, 2009 at 4:53 PM, Andrew Beekhof <beek...@gmail.com> wrote:

> On Fri, Mar 13, 2009 at 05:09, Romi Verma <romi3rd...@gmail.com> wrote:
> > Thanks for reply Dejan,
> >>
> >> No, there is no coordination between nodes. All of them will try
> >> to reset the node.
> >
> > if All of them will try then dont you think it can lead to multiple
> reset?
> > it's not good right??
>
> Far better than corrupted data.
>
> _______________________________________________
> Pacemaker mailing list
> Pacemaker@oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>
_______________________________________________
Pacemaker mailing list
Pacemaker@oss.clusterlabs.org
http://oss.clusterlabs.org/mailman/listinfo/pacemaker

Reply via email to