Failover did not work after Active node locked up

[snrgiallo]

SQL2000 (Sp2) on Win2k (SP2) cluster: 2 nodes in active/passive. The DB is backend - front ends are load
balanced IIS boxes on a content switch.
Active node locked up - no response from keyboard, no message on screen just
desktop with hourglass !

Analysis:

From system event log on the passive node , Passive node detects no heartbeat from Active node and initiates failover. It cannot assume ownership of the virtual IP addreses of the cluster or the SQL virtual server IP
address - complains that these addresses are already in use on the network. We verified that passive node
correctly assumed ownership of shared disk, quorum drive etc, but obviously due to dependencies on IP, could not start SQL on passive node and the cluster was thererfore
down. Nothing useful was seen in logs on Active node.

This was resolved by rebooting both nodes. I presume once the active node was rebooted, it released the IP addresses, as the passive node was then able to assume
ownership of IP and complete failover. Cluster was then operational.

Has anyone got any ideas on why the Active node did not release the IP address

Cheers,
S

 

[sguslan]

This may be related to the network setup on the failover machine. Check that in Network Connections-> Advanced settings, the Public network card[not the heart beat card] is shown first, if not then move it up the list.