none
Procedure for shuting down both nodes in a W2K8R2 failover cluster

    Question

  • All,

    I'm looking for the definitive guide on shutting down my 2-node windows 2008 R2 Failover cluster completely (cluster is running some file server resources).
    I've done some googling and looking at the MS help files but can't find anything which fully answers my question - maybe I haven't looked in the right places but I would have thought there would be more official advice available from MS .
    The shutdown is unfortunately necessary due to some emergency power problems which have been sprung on me with very little notice - both my SAN storage and both cluster nodes will need to go down.

    This is my shutdown plan:

     - Move all Resource groups ('services and applications' in the new terminology) to one node
     - Shut the other 'empty' node down
     - Take all 'services and applications' offline using the Failover cluster manager
     - Do a windows shutdown of the second node

    Is this sufficient or should I be doing something cluster-specific on the second node before shutting it down?

    The next step will be to shutdown my storage (NetApp)

    My startup plan is as follows:

     - Bring my SAN storage back up and confirm availability.
     - Bring up the node which was last to go down.
     - Wait for the Resource groups to come back online automatically
     - Bring up the other node
     - Test failover

    Am I missing anything?
    Would it be prudent to stop Resource groups from auto starting so that I have verified that all storage is visible first (or is this a bit paranoid?)

    Any help or advice or pointers to official documentation on this specific subject would be greatly appreciated

    Thanks,

    Tuesday, November 22, 2011 2:51 PM

Answers

  • Yes, sorry.

    This is feasible for any clustered resource.

    Your script is good:

    1- Move all your resources to single node

    2- shutdown you second node

    3- Take all your resources offline

    4- Shutdown your online node

    5- make your maintenance :)

    6- Start one of the nodes

    7- Verify that all resources are online

    8- Start your second node

    9- move some clustered resources to the second node (To optimize charge between them)

     

    Tip: For the 9th step, you can make this automatic by choosing the preferred node for your resources. With cluster feature, you can choose for a particular resource, what node or nodes that are preferred. In other terms, you can choose that some resources are preferred in the node1 and some other resources are preferred in the other node2. When only node 1 is up, all resources are brought online in the node1. When the node 2 become online, resources that prefer node 2 will automatically move to the second node.

     

     

     


    Regards, Samir Farhat Infrastructure Consultant
    Tuesday, November 22, 2011 3:37 PM
  • Yes, your script seems to be good and logic. you can proceed.

    To verify that ressources are online again to you first started Hyper-V server, you can use some cluster commands to directly see the result without need of browsing all the cluster views. 


    Regards, Samir Farhat Infrastructure Consultant
    Tuesday, November 22, 2011 3:07 PM

All replies

  • Yes, your script seems to be good and logic. you can proceed.

    To verify that ressources are online again to you first started Hyper-V server, you can use some cluster commands to directly see the result without need of browsing all the cluster views. 


    Regards, Samir Farhat Infrastructure Consultant
    Tuesday, November 22, 2011 3:07 PM
  • Thanks Samir,

    I'm not sure I've completely understood, but just to clarify, this cluster doesn't host any VMs, just some file server resources (i.e. shared folders)

    Regards,

    Patrick

    Tuesday, November 22, 2011 3:23 PM
  • Yes, sorry.

    This is feasible for any clustered resource.

    Your script is good:

    1- Move all your resources to single node

    2- shutdown you second node

    3- Take all your resources offline

    4- Shutdown your online node

    5- make your maintenance :)

    6- Start one of the nodes

    7- Verify that all resources are online

    8- Start your second node

    9- move some clustered resources to the second node (To optimize charge between them)

     

    Tip: For the 9th step, you can make this automatic by choosing the preferred node for your resources. With cluster feature, you can choose for a particular resource, what node or nodes that are preferred. In other terms, you can choose that some resources are preferred in the node1 and some other resources are preferred in the other node2. When only node 1 is up, all resources are brought online in the node1. When the node 2 become online, resources that prefer node 2 will automatically move to the second node.

     

     

     


    Regards, Samir Farhat Infrastructure Consultant
    Tuesday, November 22, 2011 3:37 PM
  • thanks Samir

    Regards,

    Patrick

    Tuesday, November 22, 2011 4:44 PM
  • Remember you need your external/ physical DC from the cluster to bring the CSV back online?  assuming you are using a CSV.
    Tuesday, November 22, 2011 11:04 PM