Restoring the cluster configuration Backing up and restoring the cluster configuration Deleting a backup configuration file

This topic describes the overall procedure for restoring a cluster configuration and the conditions that must be satisfied in order to perform a successful restore.

You can restore a cluster configuration by using the Restoring a Cluster Configuration panel in the master console or the svcconfig command in the command-line interface (CLI).

Prerequisites

Ensure that the cluster to which you are restoring the backup configuration files meets the following conditions:

In addition, the SAN Volume Controller analyzes the backup configuration file and the cluster to see if all of the following conditions are true:

Context

Hardware recovery is complete. In particular, the following hardware is operational: hosts, SAN Volume Controllers, disk controller systems, disks, and the SAN fabric.

Steps

Perform the following steps to restore your cluster configuration

  1. Create a replacement cluster, using the same SAN Volume Controllers that were present before the loss of your cluster configuration:
    1. Use service mode to delete each existing node from the cluster.
    2. Use service mode to make a new cluster on the original configuration node. Ensure that the new cluster has the same name as the one to be restored.
  2. Restore the SSH public keys.
  3. If the backup configuration files are on tertiary storage or at some other safe location, copy the file to the master console.
  4. Copy the backup configuration file from the tertiary storage to the cluster. This allows the svcconfig command to use this file during the restoration of the cluster configuration.
  5. Prepare the cluster using the svcconfig -prepare command. After issuing this command, do not rename the objects in the cluster. The default names are required to successfully complete the restoration. The unique names in the backup configuration file will be restored at the completion of the restoration.
  6. Restore the cluster configuration using the svcconfig -execute command.
Example

Select the set of backup files you wish the cluster to be recovered to. Use the master console to create a new cluster with the same name as the one to be recovered. Once the new cluster has been created, you can proceed with the restore. Perform the following steps:

  1. Log onto the cluster and clear the backup directory of any old backup restore files. Issue the following:
    ssh -l admin your_cluster_name -p 22

    svcconfig clear -all

    exit
  2. Copy the backup file from your cluster storage by issuing the following:
    scp -P 22 /offclusterstorage/svc.config.backup.xml_myconfignode
    admin@your_cluster_name:/tmp/svc.config.backup.xml

    Log back unto the cluster.
    ssh -l admin your_cluster_name -p 22
  3. Issue the following:
    svcinfo lsnode
    Ensure that only one node is online and identify which one it is. If this node was not the configuration node in the configuration you are trying to restore, either make a cluster where it is or select the appropriate backup file.
    id .... status IO_group_id IO_group_name config_node
    1 .... online 0 io_grp0 yes
  4. Issue the following:
    svcinfo lscluster
    Ensure that the cluster has been created with the same name as that in the backup files. Output similar to the following is displayed:
    Id name location
    0000020066206BE2 your_cluster_name local
  5. Issue the following:
    svcconfig restore -prepare
    This will do a comparison of the current cluster configuration and available resource and the backup file you have put onto the cluster. If there are any errors, the command will fail with a CMMVCnnnnE error. You will need to fix the error and issue the command again.
    Note:Note: If there has been any change to the fabric since the backup was taken, it will not be possible to restore the chosen configuration.

    When this command has completed, you may have received a number of warning messages. You will need to ensure that the action about to be taken is acceptable. You may need to exit the cluster and copy this log file off the cluster for reading, as it may be quite large.

  6. In order to read the log file produced by the -prepare flag to ensure you are aware of all the warnings that have been issued, exit from the cluster. Issue the following:
    exit

    scp -P 22 admin@your_cluster_name:/tmp/svc.config.prepare.log
    /offclusterstorage


    cat /offclusterstorage/svc.config.prepare.log|more
    to exit the list.
  7. When you are satisfied that the restore will happen as expected, log back onto the cluster and execute the restore command. Issue the following:
    ssh -l admin your_cluster_name -p 22

    svcconfig restore -execute
    This will use the svc.config.restore.sh file to attempt to recover your cluster structure onto the available cluster hardware. Once this has completed, you can check the log file to ensure that no errors or unexpected warnings have been issued about the restore. The following output displays that a successful restore has taken place and that no errors are reported.
    ..........................
    IBM_2145:admin>
    When you have verified the cluster as correct, you may restore your company data from the storage back onto the presented VDisks.
Post-processing requirements

Remove any unwanted configuration backup files from the cluster using the svcconfig clear command.

Related topics Backing up and restoring the cluster configuration Deleting a backup configuration file

(C) Copyright International Business Machines Corporation 2003, 2004