IBM Books

Managing Shared Disks

Replacing a server

If you need to relocate a virtual shared disk server and its secondary server to another node, do the following:

  1. Archive the System Data Repository by entering SDRArchive on the command line. If you make a mistake or change your mind, this will permit you to undo the changes you make.
  2. Bring down any applications that use the affected servers.
  3. Stop the RVSD subsystem on the affected servers. For example:

    ha.vsd stop

    If you stop RVSD on all the nodes you can skip the next step.
  4. Suspend and stop all virtual shared disks involved on all the nodes. For example:
    suspendvsd vsd_name
    stopvsd vsd_name
  5. Unconfigure all virtual shared disks involved on all the nodes. For example:
    ucfgvsd vsdname
  6. If you need to remove or alter any of the original nodes, you can do so now. Refer to PSSP: Installation and Migration Guide for more information.
  7. Recable the disks to attach them to the new servers.
  8. Set up the volume groups on the new server nodes. For example:
    importvg -y volume_group_name hdiskx
    chvg -a n volume_group_name
    varyoffvg volume_group_name
    vsdvgts -a volume_group_name
  9. Designate the new servers as virtual shared disk nodes. See Chapter 3, Understanding your Managing Shared Disks process for more information.
  10. Change the System Data Repository to reflect the new nodes for the affected virtual shared disk volume group. From any node in the partition, or from the control workstation, issue the SDRGetObjects command to see the current settings.
  11. Issue a reset to make all the virtual shared disks in the configuration active and to start the RVSD subsystem on the new servers. For example:
    ha_vsd reset
  12. Refresh the RVSD subsystem so that all the nodes will become aware of the changed servers. This only needs to be done from one node where the RVSD subsystem is active, since it runs a protocol to all active RVSD subsystems. For example:
    ha.vsd refresh
  13. To verify that the driver has a current list of the existing virtual shared disk nodes, issue lsvsd -i. Do this on every node before attempting any remote virtual shared disk requests to the node.

[ Top of Page | Previous Page | Next Page | Table of Contents | Index ]