Top
PRIMECLUSTER  Installation and Administration Guide4.3

13.5.4 Recovery Operation for When an Error Occurs on the Non-Global Zone

If using warm-standby, follow "7.4 Corrective Actions for Resource Failures" and perform the normal recovery operation.

In the following conditions, a non-global stop process will be performed using the Cmdline resource set up to the global zone if an error occurs to a non-global zone, or the RMS or applications running on the non-global zone:

When performing a non-global stop process by the Cmdline resource, you cannot log into the non-global zone and perform recovery operations.

The following shows the troubleshooting procedure for such a case.

Note

Before performing this troubleshooting procedure, check that the corresponding cluster application is in a stop status on other nodes.

  1. Clearing the Faulted status on the global zone

    Clear the Faulted status for the global zone cluster application. For the procedure using GUI, refer to "7.2.2.4 Bringing Faulted Cluster Application to Online State." If using CLI, execute hvutil -c <cluster application name>.

  2. Migrating to maintenance mode

    From the global zone, migrate the corresponding cluster application to maintenance mode. It is possible to use GUI and CLI. For the procedure, refer to "7.3 Using maintenance mode" of the "RMS with Wizard Tools Configuration and Administration Guide."

  3. Activating the Gds resource and Fsystem resource

    Manually perform the mounting of the file system and the startup of GDS volume necessary to the startup of the non-global zone. For the startup procedure for GDS volume, refer to "5.1.10 Operations from GDS Management View" of the "PRIMECLUSTER Global Disk Services Configuration and Administration Guide." Mount the file system using the mount command.

    This step is not necessary in the case of not sharing non-global zone images or a single-node cluster is operated.

  4. Attaching and starting up the non-global zone

    After stopping the cluster application, the non-global zone enters a detached status, so perform the attaching and starting up of the non-global zone using zoneadm attach and zoneadm boot.

    # zoneadm -z zone-a attach -F   *1
    # /usr/lib/brand/solaris8/s8_p2v zone-a (
    For Solaris 8 Container) *2
    # /usr/lib/brand/solaris9/s9_p2v zone-a (For Solaris 9 Container) *2
    # zoneadm -z zone-a boot *2

    *1: Perform this only if sharing a non-global zone between cluster nodes.

    *2: "zone-a" is the zone name.

  5. Performing the maintenance operation in the non-global zone

    Perform the necessary maintenance operations on the non-global zone. One is permitted to use RMS commands (hvcm, hvshut, hvswitch, hvutil, hvdisp, etc.) as needed.

    Note

    If something such as a shared disk for data takeover is having maintenance performed using a resource set up to the RMS, be sure that the RMS for the non-global zones allocated to each of the global zones are not simultaneously Online.

  6. Stopping and detaching the non-global zone

    Return the non-global zone to a detached status. From the non-global zone, after executing such commands as "shutdown -y -g0 -i0" and stopping the non-global zone, detach the non-global zone using zoneadm detach.

    # shutdown -y -g0 -i0
    # zoneadm -z zone-a detach
  7. Deactivating the Gds resource and Fsystem resource

    Return the resources activated with the above Step 3 "Activating the Gds resource and Fsystem resource" back to a deactivated status.

    Manually perform an unmount of the file system and a stopping of the GDS volume. Unmount the file system using the unmount command. For the procedure on stopping the GDS volume, refer to "5.1.10 Operations from GDS Management View" of the "PRIMECLUSTER Global Disk Services Configuration and Administration Guide."

    If the Gls resource was also activated, manually perform deactivation.

    This step is not necessary in the case of not sharing non-global zone images or a single-node cluster is operated.

  8. Canceling the maintenance mode

    From the global zone, remove the corresponding cluster application from maintenance mode. It is possible to use GUI and CLI. For the procedure, refer to "7.3 Using maintenance mode" of the "RMS with Wizard Tools Configuration and Administration Guide."

  9. Starting up the cluster application

    Start up the cluster application from the global zone as needed.