PRIMECLUSTER Installation and Administration Guide4.6 Cloud Services

6.3.2 Procedure for Restoring Nodes While the Operation does not Work

1. If either or all of the nodes are started before restoring the nodes, stop RMS. Perform the restore procedure on either node that is started.

   # hvshut -a

2. Select the latest disk when all nodes are started before restoring nodes in an environment where the mirroring among the servers of GDS is used. For all classes of GDS, execute the following command on either node.

   # /etc/opt/FJSVsdx/bin/sdxnetdisk -S -c <class name>

3. Refer to "6.3.3 Restoring the Virtual Server from the Snapshot" to restore one node and then start the node.

   Note

   If you did not set the expanded volume used in the mirroring among the servers of GDS according to "6.3.3 Restoring the Virtual Server from the Snapshot", refer to "6.3.3 Restoring the Virtual Server from the Snapshot" to restore the virtual server again without attaching the target volume to the restored virtual server. If the target volume is attached to the restored virtual server, the process fails in the rest of the procedure.

4. When using the mirroring among the servers of GDS, start the node and then execute the following with the restored node.

   1. Delete the information of the iSCSI device.

      # rm -f /var/opt/FJSVsdx/log/.sdxnetmirror_disable.db
      # rm -f /var/opt/FJSVsdx/log/.sdxnetmirror_timestamp

   2. Stop RMS.

      # hvshut -l

5. When restoring the other node, refer to "6.3.3 Restoring the Virtual Server from the Snapshot" to restore the node.

   Note

   If the volume used in the mirroring among the servers of GDS is attached after creating a virtual server, the process fails in the rest of the procedure. Refer to "6.3.3 Restoring the Virtual Server from the Snapshot" to create the virtual server again.

6. When using the mirroring among the servers of GDS, delete the iSCSI device information with the restored node if the node is restored with step 5.

   # rm -f /var/opt/FJSVsdx/log/.sdxnetmirror_disable.db
   # rm -f /var/opt/FJSVsdx/log/.sdxnetmirror_timestamp

7. If all nodes are stopped before restoring nodes in an environment where the mirroring among the servers of GDS is used, check the status of the source slice for the synchronization copying. If the source slice for the synchronization copying is INVALID, restore the status of the slice. For the "-d" option of the "sdxfix " command, specify the source disk of the synchronization copying. Perform this procedure on either node.

   # sdxfix -V -c <class name> -v <volume name> -d <disk name> -x NoRdchk

8. If 2 of step 4 is performed, start RMS on the node where RMS is stopped.

   # hvcm

Note

If the virtual server name is changed in step 3 or step 5, perform step 3 to 5 in "3.7.1.2 Setting up the Shutdown Facility" on all nodes.