the kirschi

Hi all,

I setup an active-/passive two-node Exchange cluster (2003 R2) and asked
myself how to configure the cluster so it will failover in the event of a
specific failure.

So the question is what events should be configured to cause a failover and
how to do this.

One example:
I expected the cluster to fail over if LAN connectivity is lost on one node,
but when I disconnect the cable manually nothing happens. So how do I
configure the cluster so that the heartbeat connection recognizes the network
disconnection and initializes a failover.

Thanks for your help,
The Kirschi

chuck timon microsoft

I am sure you do not want to totally disrupt your cluster for a heartbeat
network failure. It would be different if it were a Public network because
now services are now longer available. Unplugging one of the heartbeat
cables just forces the cluster to use the Public network for internal
communications and that is why we tell you to configure the Public Network
for "All communications" for just this scenario.

Now, if you test by pulling the Public cable on one of the nodes and that
node owns an EVS resource....you will be surprised at the behavior. You
will see your EVS group(s) take about 10-15 minutes to actually failover.
This is because of the behavior of the Exchange cluster resource not
immediately reacting to the failure of a public network. The cluster
service recognizes it immediately and passes the info to exres.dll, but
Exchange 'delays' it's response. Unfortunately, this is the behavior in
Exchange 2000 and 2003. I don't know what the behavior will be in the next
version of Exchange. There is no 'tweak' to make it faster. This behavior
does not happen for SQL or file shares....just Exchange. Spooler resources
take a long time because of the 'spooled' jobs that are printing.

--
Chuck Timon, Jr.
Microsoft Corporation
Longhorn Readiness Team
This posting is provided "AS IS" with no
warranties, and confers no rights.