We have a four node NW6.5 SP3 cluster that uses our production network for its
heartbeat (nodes have a single NIC). We recently added a second supervisor
module to our core Cisco router/switch to which the nodes connect directly (and

While testing failover from one supervisor to the other (which takes about 10
seconds), we noticed that the cluster will nonetheless abend. (Our heartbeat
threshold is 16 seconds.)

Does anybody have switch/router redundancy like this setup where the cluster can
tolerate a failover? Any tips?

I know this is probably a good argument for a separate heartbeat network,
however, it seems that most folks try to avoid these, and we have a lot of
redunancy (dual power supplies, etc.) on our core switch which we would have to
replicate for a heartbeat network.


Tony Skalski
Systems Administrator
St. Olaf College 1510 St. Olaf Avenue
Information and Northfield, MN
Instructional Technologies 55057-1097
================================================== =====