Windows 2003 Cluster: Failover Delay

Posted by Ramon Marco Navarro on Server Fault See other posts from Server Fault or by Ramon Marco Navarro
Published on 2010-03-28T03:41:31Z Indexed on 2010/03/28 3:43 UTC
Read the original article Hit count: 503

I am testing the failover policies of our test failover cluster system. When I shutdown the node who is currently controlling the cluster (NODE1), it takes about 2 mins and 40 seconds before the next node on the preferred list (NODE2) takes control of the cluster.

I tried changing the looksAlive and isAlive interval to 5000ms to all resources, but that didn't help. Looking at the Event Viewer of the remaining nodes, it shows that it was almost instantly detected that NODE1 was down. But it took another ~2:40 minutes for it to be removed from the live cluster list and for NODE2 to take over.

Is there anyway of changing or shortening this "failover delay"? This is the setup of the cluster:

  • (1) One ClusterDC connected to the public network
  • (3) Three nodes running Win2003 with a quorum type of MNS
  • Private network is connected to network hub
                   ________________                       _________________
(ClusterDC)=------=|              |=------=(Node1)=------=|               |  
                   |Public Network|=------=(Node2)=------=|Private Network|
                   |   (Switch)   |=------=(Node3)=------=|     (Hub)     |
                   ----------------                       -----------------

© Server Fault or respective owner

Related posts about windows-server-2003

Related posts about windows-cluster