Does anyone know how to prevent a cluster group from starting. I have a
situation whereby I have several standby groups that have resources that are
not accessible without manual intervention which act as standby groups and
so need to remain offline until DR situation is invoked (there are also live
groups on this server too though). Therefore I want to prevent people from
manually accidently starting them and would like to know the best way to do
that. Secondly because these groups *have* been accidently started (and
failed) I would like to reset there status if possible so esthetically it
looks healthy.

Finally if someone can answer those two things a gold star will be given for
the third and final question. Is it possible to forcefully stop a group
which is failing back and forth?

Many thanks,

Mark.

If your group is in an offline state, it will stay in an offline state until
someone manually attempts to bring it online. So just leaving it in an
offline state could work. Though if your group is online at the time of the
failover, this wouldn't work. It also wouldn't prevent anyone from bringing
it online accidentally.

A better way would be to change the "Possible Owner" value of one (or all)
of the resources within this group to remove the DR node. That way, the
group won't be allowed to move to that other node because the resources
would not be allowed to come online on that node. One would need to manually
add back the DR node as a possible owner before you would be allowed to
bring this group online on the node.

Forcefully stopping a group which is failing back and forth is sometimes
challenging depending upon which resources are causing the ping-ponging to
occur. If you have a specific resource that is causing the failover, one way
to top it from failing over would be to change the restart values of the
resource. By default, all resources are set to restart and to affect the
entire group when the resource fails. You could adjust the value to either
make the failing resource not attempt to restart, or you could just uncheck
the "Affect the group" box so that when the resource fails, the it would
just fail and not cause the group to failover.

If you've got a stubborn resource that is stuck in an Online Pending state,
you can try to issue a "cluster res "resource name" /fail" command to try to
make it fail. If that doesn't work, you would potentially need to kill
cluster services.

Hope this helps.

Regards,
John

Visit my blog: http://msmvps.com/blogs/jtoner

"Mark Broadbent" <nospam (AT) nospam (DOT) com> wrote