How to properly shut down Windows Server 2008 R2 cluster?

windows-server-2008-r2 cluster failovercluster

10,612

Those two "options" accomplish completely different things.

If you pick Shut down Cluster..., this will stop ALL services hosted by the cluster and stop the cluster service on all of the servers. None of the servers will actually shut down.

If you shut down Windows, the resources hosted on that server will be moved to another member of the cluster. When that is complete, then the server will shut down normally.

If you are ultimately shutting down all servers in a cluster, they will eventually accomplish the same thing. My recommendation (if you really want to shut down all the servers in the cluster) is to use Shut down Cluster.... That way all of the resources go down right away and you don't have to wait for them to hop around.

10,612

Volodymyr Molodets

Updated on September 18, 2022

Comments

Volodymyr Molodets almost 2 years

We need to shut down MS Failover Cluster running on two nodes with MS Windows Server 2008 R2 Enterprise for hardware maintenance.

I recently came across two possible approaches on how to shut down the cluster:

First approach says to take the cluster resources offline and proceed to windows shutdown, on the other hand if you right-click on cluster name in FC snap-in under More Actions... you get Shut down Cluster...

What's the difference between these two approaches? Which is the recommended way to shut down the cluster? Does it matter what kind of resource is running on cluster, i.e. MS SQL, Hyper-V, custom application?

This is active/passive cluster.
- longneck over 11 years
  
  Are you trying to shut down all servers in the cluster? Or do you just need to shut down one server at a time?
- Volodymyr Molodets over 11 years
  
  @longneck - all servers(i.e. nodes) in the cluster.There are totally two nodes.
longneck over 11 years

the behaviors are documented.
Greg Askew over 11 years

If shutting down the servers, probably best to start with the passive node(s) first, to prevent unnecessary failover.
longneck over 11 years

@GregAskew Provided there is a passive node. For example a Hyper-V cluster, all of the hosts are likely hosting active VM's. That's why I suggested using Shut down Cluster....