none
Connecting new node to iSCSI SAN NTFS Error

    Domanda

  • We have a two node Hyper-V Server R2 cluster and are trying to connect a third node to the cluster but when I try and make the connections to our iSCSI SAN the system hangs when trying to mount the volumes.  Then when I look on the event viewer for the node there are a bunch of errors:

    Source: NTFS, Error: "The file system structure on the disk i scorrupt and unusable.  Please run the chkdsk utility on the volume X:"

    Then when I try to go in and look at the Disk Management on the new node it won't load.

    The cluster is running just fine but I can't get the new node to see the LUNs on the SAN.

     

    lunedì 25 ottobre 2010 17:49

Risposte

  • I was able to get the third node added.

    Configure the server to the point where you want to attach it to the SAN then make the connections to the SAN but do not try to add the devices via the iSCSI Intiator and do not try to connect to the disks via the Disk Management console.  Just make sure that the connections status is Connected.

    Now add the node to the cluster.  After is has been added to the cluster you will be able to get to the LUNs through the explorer (C:\ClusterStorage) but you are still not ready to browse to the disks through the Disk Management or add the devices through the iSCSI Initiator.

    Now change the quorum settings to the recommended for three nodes (node-majority) this will release the disk that was being used for the quorum making it available storage.  Now go into the storage section and offline that disk.  Do not remove the drive because if you go back to an even number of nodes you'll probably need the quorum/witness disk again, just leave it offline.

    Now go back into the iSCSI Initiator and add the devices and you can now go into the Disk Management and view the drives.

    That was my experience...if someone can verify the steps and/or try it and let me know how it worked for you.

    domenica 31 ottobre 2010 02:53

Tutte le risposte

  •  

    Hi,

    my 1st question would be here.....did you had run the validation wizard before you add the node? and what are your exactly steps when you had the node?

    1. network configuration + FC role

    2. modify SAN to present shared disk also to 3rd node (+ if ISCSI used here, configure ISCSI client)

    2. start FC manager and validate 3rd node

    3. add 3rd node to cluster

    if this fails, some more information about eventlogs generated by Failover cluster/system would be good. if there is nothing useful you can also generate an cluster log (cluster log /gen).

    "....Then when I try to go in and look at the Disk Management on the new node it won't load...."

    when you present the disk to the new node (ISCSI/FC) you should initate an rescan from disk management but latest after an reboot the disk MUST be in disk management otherwise there is something wrong with your zoning/ISCSI authentification.

    Also here are links around quorum model when you add nodes to your cluster:

    Configuring the Quorum in a Failover Cluster

    http://technet.microsoft.com/en-us/library/cc770620%28WS.10%29.aspx#BKMK_requirements

    How quorum works in failover cluster:

    http://technet.microsoft.com/en-us/library/cc730649%28WS.10%29.aspx

    Hope that helps.

    Regards

    Ramazan


    Ramazan Can [MVP Cluster] http://ramazancan.wordpress.com/
    • Modificato Ramazan Can lunedì 25 ottobre 2010 18:52 formatting
    lunedì 25 ottobre 2010 18:50
  • To clarify I haven't tried to join the server to the cluster yet, I'm running into an issue with the configuration/disk prep to try and join.

    Steps:

    1. Installed Hyper-V Server R2
    2. Configured server and network adapters (install updates, date/time, join domain, assign IP addresses, etc...)
    3. Install iSCSI Initiator, MPIO, and Failover Cluster
    4. Configure iSCSI connections to SAN

    and this is where I'm running into the problem.  I can connect to the SAN and create the multi-paths just fine but when I try to configure the SAN devices the iSCSI configurator hangs.  Then when I go the Hyper-V server and look in the event viewer that's where I'm seeing the Ntfs disk corrupt errors, and then when I try to look at the Disk Management to see if it saw the LUNs, it hangs as well.

    There are already two servers in the cluster and they are connected up to the SAN just fine.  Everything is configured the same.  It seems like NTFS isn't allowing the new server to view the LUNs.

     

    lunedì 25 ottobre 2010 19:12
  • Hi,

     

    As you mentioned that you were running Hyper-V Server R2, so I would like to know whether you dedicate one LUN for one VM or set Cluster Shared Volumes on the cluster.

     

    By the way, you can check the following post.

     

    http://social.technet.microsoft.com/Forums/en-US/winserverhyperv/thread/86ba2313-8bd8-42fd-adfd-b89af38695fe

     

     

    Best Regards,

    Vincent Hu

     

    mercoledì 27 ottobre 2010 08:54
    Moderatore
  • It is Cluster Shared Volumes.  There are five LUNs with multiple VMs per LUN and a 6th LUN for the quorum.

    Am I supposed to connect to the SAN and then just join it to the cluster without connecting to the devices in the iSCSI initiator and accessing Disk Management on the host before hand?

    If I do when I join it to the cluster will it connect the devices in the iSCSI initiator and add it to CSV?

    giovedì 28 ottobre 2010 00:04
  • Hi,

    I'm having a similar problem as beauak too...

    node1 and node2 are running Hyper-V Server R2 in the cluster. They are sharing one LUN with enabled Cluster Shared Volumes. "Failover Cluster Validation Report" returns all green on existing cluster.

    Disks are connected as iSCSI targets and served by Microsoft Storage Server 2008.

    I want to add 3'd node to the cluster. Before connecting i need to configure disks on node3 to match cluster configuration; ON iSCSI target no problems to create new target and attach the same Virtual Disks. When configuring iSCSI Target on node3 (iscsicpl) after adding Target and invoking "Volumes and Devices\Auto Configure" screen hangs there for ~10min but finally configures disks. To connect to the node3 via "Computer Manager\Disk Management" takes about the same time and usually times out before successfully connecting. Event viewer shows the Ntfs disk corrupt errors.

    NOTE: iscsicpl "Volumes and Devices\Auto Configure" and “Computer Manager\Disk Management” on nodes in the cluster (node1 and node2) runs without an issue and response is instant.

    giovedì 28 ottobre 2010 00:12
  • I was able to get the third node added.

    Configure the server to the point where you want to attach it to the SAN then make the connections to the SAN but do not try to add the devices via the iSCSI Intiator and do not try to connect to the disks via the Disk Management console.  Just make sure that the connections status is Connected.

    Now add the node to the cluster.  After is has been added to the cluster you will be able to get to the LUNs through the explorer (C:\ClusterStorage) but you are still not ready to browse to the disks through the Disk Management or add the devices through the iSCSI Initiator.

    Now change the quorum settings to the recommended for three nodes (node-majority) this will release the disk that was being used for the quorum making it available storage.  Now go into the storage section and offline that disk.  Do not remove the drive because if you go back to an even number of nodes you'll probably need the quorum/witness disk again, just leave it offline.

    Now go back into the iSCSI Initiator and add the devices and you can now go into the Disk Management and view the drives.

    That was my experience...if someone can verify the steps and/or try it and let me know how it worked for you.

    domenica 31 ottobre 2010 02:53
  • Can you explain in more detail this process?

    What do you mean by "make the connections to the SAN but do not try to add the devices via the iSCSI initiator"?

    Do you add the node to the cluster even though it doesn't pass the validation? my understanding is that if the drives aren't mounted with matching drive letter assignments it won't pass the validator?

    giovedì 17 gennaio 2013 18:38