Failoverclusteringproblems on one Hyper-V-Server with Hyper-V Clients

Answered Failoverclusteringproblems on one Hyper-V-Server with Hyper-V Clients

  • Thursday, February 21, 2013 2:44 PM
     
     

    Hi,

    I hope this is the right Forum to post my problem, because it is a little bit mixed up with hyperv and windows failoverclustering.

    My problem is the following:
    I have three physical server, that host some hyper-v clients. Lets call them srv01, srv02 and srv03. All three are running Windows Server 2012 Datacenter with all updates and actual drivers. All Servers are Dell PowerEdge R710. srv02 and srv03 are absolute identical except that srv02 has 288GB RAM and srv03 has 144GB RAM. srv01 has a different processor and motherboard. RAM is 288GB too.

    I installed a Windows Server 2012 Datacenter on srv01 and srv02 as a hyper-v client. Lets call them hv01 and hv02. I installed all available updates via Windows Update. After this, I tried to create a Failovercluster with using the windows-feature. While creating the cluster from hv01 with hv01 and hv02 as a node, the creationprocess failed with a timeout to hv02. When I try to create the cluster from hv02, the process timed out to hv01. When I created the cluster with just one node, the creationprocess finished successful. Then I tried to add the second node to the existing one-node-cluster, but it failed again with a timeout.

    After two days of searching analysing etc. I moved hv02 to srv01. So hv01 and hv02 are running on srv01. After doing this, the clustercreation with both nodes was successful. Now I moved hv02 back to srv02. That results in a not available clusternode - so hv02 was not available in the failovercluster.

    Then I moved hv02 to srv03. And voila: the cluster was up. I tested some configurations and figured the following out:
    hv01 and hv02 on srv01 - works
    hv01 on srv01 and hv02 on srv02 - fails
    hv01 on srv01 and hv02 on srv03 - works
    hv01 on srv02 and hv02 on srv03 - fails

    After this, I was a little bit confused, because I thought it might be the different hardware, but that doesn't make sense, because srv02 and srv03 are identical and srv02 is not workling.

    So: Anyone has an idea what the problem can be?


All Replies

  • Thursday, February 21, 2013 4:13 PM
     
     
    Have you run a full cluster validation? That should give you exactly where the problem is (or at least guide you in the right direction).
  • Thursday, February 21, 2013 6:12 PM
     
     

    Hi Ted,

    I ran a check and all points are successfull. Before creating the cluster with two nodes and before trying to add the second node to the one-node-cluster.

    So everything seems to be fine. The same machine on another Server was fine after Live-Migrating the machine to another server...

  • Monday, February 25, 2013 7:26 AM
     
     

    So does everything work now, or does it ONLY work on srv01/srv01 and srv01/srv03? Isn't completely clear for me right now.

  • Monday, February 25, 2013 8:33 AM
     
     

    Hi Jamilian,

    currently it is not working.

    As you mentioned it only works on srv01/srv01, srv01/srv03 and srv03/srv03. srv02/srv02 is also working!
    It is not working on srv01/srv02 and srv02/srv03

  • Wednesday, May 15, 2013 8:11 AM
     
     Answered

    Hi @all

    The problem is now solved.
    Even the Microsoft Support didn't find any reason for this.

    With try and error we figured out (without MS), that the problem is gone when we disable the "Microsoft Failover Cluster Virtual Adapter Performance Filter" in the settings for the networkcard of srv02. In srv01/srv03 the setting was enabled but is now disabled too.

    BTW: This happend in another environment too. So it is not an error, that just occures once in our environment.

    • Marked As Answer by Hunv Wednesday, May 15, 2013 8:11 AM
    •