none
Quorum Disk failure RRS feed

  • Question

  • We have our quorum disk that failed. We want to replace it with a new disk. Our san is older and will have to swap the old disk with a new disk. Should there be a process to make sure after we replace the old with new disk we should follow?
    Friday, April 9, 2010 5:40 PM

Answers

  • Take a look at these articles:

    http://support.microsoft.com/kb/305793

    http://support.microsoft.com/default.aspx/kb/280353?p=1

     


    Santhosh Sivarajan | MCTS, MCSE (W2K3/W2K/NT4), MCSA (W2K3/W2K/MSG), CCNA, Network+ Houston, TX http://blogs.sivarajan.com/ http://publications.sivarajan.com/ This posting is provided "AS IS" with no warranties, and confers no rights.
    Friday, April 9, 2010 5:45 PM
  • Hi 562networks

    I am assuming this is win2k3 cluster and not win2k8.

    The cluster recovery utility allows a new disk, managed by a new physical disk resource to be substituted in the resource dependency tree and for the old disk resource (which now no longer has a disk associated with it) to be removed.

    To replace a failed disk use the following procedure:

    • Add a new disk drive to the cluster. In a storage area network environment, adding a new disk drive may involve creating a new logical unit and exposing it to the server cluster nodes with appropriate LUN masking, security and zoning properties.
    • Make sure that the new disk is only visible to one node in the cluster. Until the Cluster service takes control of the new disk and a physical disk resource is created, there is nothing to stop all nodes that can see the disk from accessing it. To avoid file system issues, you should try to avoid exposing a disk to more than one node until it has been added to the cluster. In some cases (such as with low-end fiber channel RAID devices or devices in a shared SCSI storage cabinet) there is no way to avoid multiple machines from accessing the same disk. In these cases, a CHKDSK may run when the disk resource is brought online in step 5 of this procedure. Although this situation is recoverable through CHKDSK, you can avoid it by shutting down the other cluster nodes, although this may not be appropriate if the cluster is hosting other, currently functioning applications and services.
    • Partition and format the new disk drive as required. Note: For a disk drive to be considered as a cluster-capable disk drive, it must be an MBR format disk and must contain at least one NTFS partition. Assign it a drive letter other than the letter it is replacing for now.
    • Create a new physical disk resource for the new disk drive using Cluster Administrator (or the cluster.exe command line utility).
    • Make the disk drive visible to the same set of nodes as the disk drive that it is replacing (in a typical configuration, a disk driver is visible to all nodes in the server cluster). In the event that the device does not appear to the cluster nodes, you may perform a manual rescan for new hardware using the device manager. At this stage you should try to bring the disk resource online and then fail it over all nodes of the cluster in turn to ensure that the new physical disk is correctly configured and can be viewed from all nodes.
    • Use the Server Cluster Recovery Utility to substitute the newly created physical disk resource for the failed resource. Note: The Server Cluster Recovery Utility ensures that the old and new disk resources are in the same resource group. It will take the resource group offline and transfer the properties of the old resource (such as failover policies and chkdsk settings) to the new resource. It will also rename the old resource to have "(lost)" appended to the name and rename the new resource to be the same as the old resource. Any dependencies on the old resource will be changed to point to the new resource.
    • Change the drive letter of the new physical disk to match that of the failed disk. Note: The new physical disk resource must be brought online first and then the drive letter can be changed (on the node hosting the physical disk resource) using the Disk Management snap-in available via Computer Management.
    • Once you have validated that the new resource is correctly installed, you should delete the old physical disk resource as it no longer represents a real resource on the cluster.
    • Once the cluster is configured, you should restore the application data to the new disk drive.

    Gaurav Anand
    Sunday, April 11, 2010 9:38 AM
  • Hi,

    If the Quorum disk is on hard raid you can change the disk online without broke anything

    :)

     


    Dhafer
    Monday, April 12, 2010 5:00 PM

All replies

  • Take a look at these articles:

    http://support.microsoft.com/kb/305793

    http://support.microsoft.com/default.aspx/kb/280353?p=1

     


    Santhosh Sivarajan | MCTS, MCSE (W2K3/W2K/NT4), MCSA (W2K3/W2K/MSG), CCNA, Network+ Houston, TX http://blogs.sivarajan.com/ http://publications.sivarajan.com/ This posting is provided "AS IS" with no warranties, and confers no rights.
    Friday, April 9, 2010 5:45 PM
  • Hi 562networks

    I am assuming this is win2k3 cluster and not win2k8.

    The cluster recovery utility allows a new disk, managed by a new physical disk resource to be substituted in the resource dependency tree and for the old disk resource (which now no longer has a disk associated with it) to be removed.

    To replace a failed disk use the following procedure:

    • Add a new disk drive to the cluster. In a storage area network environment, adding a new disk drive may involve creating a new logical unit and exposing it to the server cluster nodes with appropriate LUN masking, security and zoning properties.
    • Make sure that the new disk is only visible to one node in the cluster. Until the Cluster service takes control of the new disk and a physical disk resource is created, there is nothing to stop all nodes that can see the disk from accessing it. To avoid file system issues, you should try to avoid exposing a disk to more than one node until it has been added to the cluster. In some cases (such as with low-end fiber channel RAID devices or devices in a shared SCSI storage cabinet) there is no way to avoid multiple machines from accessing the same disk. In these cases, a CHKDSK may run when the disk resource is brought online in step 5 of this procedure. Although this situation is recoverable through CHKDSK, you can avoid it by shutting down the other cluster nodes, although this may not be appropriate if the cluster is hosting other, currently functioning applications and services.
    • Partition and format the new disk drive as required. Note: For a disk drive to be considered as a cluster-capable disk drive, it must be an MBR format disk and must contain at least one NTFS partition. Assign it a drive letter other than the letter it is replacing for now.
    • Create a new physical disk resource for the new disk drive using Cluster Administrator (or the cluster.exe command line utility).
    • Make the disk drive visible to the same set of nodes as the disk drive that it is replacing (in a typical configuration, a disk driver is visible to all nodes in the server cluster). In the event that the device does not appear to the cluster nodes, you may perform a manual rescan for new hardware using the device manager. At this stage you should try to bring the disk resource online and then fail it over all nodes of the cluster in turn to ensure that the new physical disk is correctly configured and can be viewed from all nodes.
    • Use the Server Cluster Recovery Utility to substitute the newly created physical disk resource for the failed resource. Note: The Server Cluster Recovery Utility ensures that the old and new disk resources are in the same resource group. It will take the resource group offline and transfer the properties of the old resource (such as failover policies and chkdsk settings) to the new resource. It will also rename the old resource to have "(lost)" appended to the name and rename the new resource to be the same as the old resource. Any dependencies on the old resource will be changed to point to the new resource.
    • Change the drive letter of the new physical disk to match that of the failed disk. Note: The new physical disk resource must be brought online first and then the drive letter can be changed (on the node hosting the physical disk resource) using the Disk Management snap-in available via Computer Management.
    • Once you have validated that the new resource is correctly installed, you should delete the old physical disk resource as it no longer represents a real resource on the cluster.
    • Once the cluster is configured, you should restore the application data to the new disk drive.

    Gaurav Anand
    Sunday, April 11, 2010 9:38 AM
  • Hi,

    If the Quorum disk is on hard raid you can change the disk online without broke anything

    :)

     


    Dhafer
    Monday, April 12, 2010 5:00 PM