Virtual machine monitoring
Ensuring high availability of services running in clustered VMs is
important because service interruptions can lead to loss of user
productivity and customer dissatisfaction. A new capability of Failover
Cluster Manager in Windows Server 2012 is the ability to monitor the
health of clustered VMs by determining whether business-critical
services are running within VMs running in clustered environments. By
enabling the host to recover from service failures in the guest, the cluster
service in the host can take remedial action when necessary in order to
ensure greater uptime for services your users or customers need.
You enable this functionality by right-clicking the clustered VM and
selecting Configure Monitoring from the More Actions menu item, as
shown here:
You then select the service or services you want to monitor on the
VM, and if the selected service fails, the VM can either be restarted
or moved to a different cluster node, depending on how the service
restart settings and cluster failover settings have been configured:
You can also use PowerShell to configure VM monitoring. For example, to configure monitoring of the Print Spooler service on the VM named SRV-A, you could use this command:
Add-ClusterVMMonitoredItem -vm SRV-A -service spooler
For VM monitoring to work, the guest and host must belong to the
same domain or to domains that have a trust relationship. In addition,
you need to enable the Virtual Machine Monitoring exception in Windows Firewall on the guest:
If PowerShell Remoting is enabled in the guest, then you don’t need
to enable the Virtual Machine Monitoring exception in Windows Firewall
when you configure VM
monitoring using PowerShell. You can enable PowerShell Remoting by
connecting to the guest, opening the PowerShell console, and running
this command:
Enable-PSRemoting
Then, to configure monitoring of the Print Spooler service on the
guest, you would open the PowerShell console on the host and run these
commands:
Enter-PSSession
Add-ClusterVMMonitoredItem -service spooler
Exit-PSSession
VM monitoring can
monitor the health of any NT Service such as the Print Spooler, IIS, or
even a server application like SQL Server. VM monitoring also requires the use of Windows Server 2012 for both the host and guest operating systems.
The quorum for a failover cluster is the number of elements that
need to be online in order for the cluster to be running. Each element
has a “vote,” and the votes of all elements determine whether the
cluster should run or cease operations. In the previous version of
Failover Clustering in Windows Server 2008 R2, the quorum could include
nodes, but each node was treated equally and assigned one vote. In
Windows Server 2012, however, the quorum settings can be configured so that some nodes in the cluster have votes (their vote has a weight of 1, which is the default), whereas others do not have votes (their vote has a weight of 0).
Node vote weights
provide flexibility that is particularly useful in multisite clustering
scenarios. By appropriately assigning a weight of 1 or 0 as the vote
for each node, you can ensure that the primary site has the majority of
votes at all times.
Another new feature of Failover Clustering in Windows Server 2012 is
the ability to change the quorum dynamically based on the number of
nodes currently in active membership in the cluster. This means that as
nodes in a cluster are shut down, the number of votes needed to reach
quorum changes instead of remaining the same, as in previous versions
of Failover Clustering.
Dynamic quorum allows a failover cluster to remain running even when
more than half of the nodes in the cluster fail. The feature works the
following quorum models:
It does not work, however, with the Disk Only quorum model.
