I have been doing some testing lately that required me to increase the failover threshold of a Windows 2008 failover cluster. The particular application was Exchange 2010 and a database availability group (DAG). Windows failover clustering gives a few options to adjust the amount of time the nodes can be without communication before a node is considered failed. By default this is 5 seconds for nodes located in the same subnet and different subnets. In the screenshots below the parameters CrossSubnetDelay and CrossSubnetThreshold are used for clusters that have nodes in different subnets. The parameters SameSubnetDelay and SameSubnetThreshold are used when nodes are in the same subnet. The delay parameters specify the amount of time (in milliseconds) that the cluster network driver waits between sending cluster service heartbeats. The threshold parameters specify how many cluster service heartbeats can be missed before it is determined that the cluster service ha stopped. The table below shows each parameter and it's default and lower and upper limits.
In the screenshots below I change the samesubnet settings on my test cluster, in this case my cluster is called TSLAB-DAG-1
- To see the clusters installed issue the command: cluster.exe /list
- To see the properties currently in use: cluster.exe tslab-dag-1 /prop
To set the new settings: cluster.exe /cluster:tslab-dag-1 /prop samesubnetdelay=2000
If your looking for a defailed paper on Windows 2008 Clustering Architecture check this out: https://download.microsoft.com/download/3/b/5/3b51a025-7522-4686-aa16-8ae2e536034d/Windows%20Server%202008%20Failover%20Clustering%20Architecture%20Overview.doc