Remove Member (Active-Active) operation

Run this operation to permanently remove an existing instance (primary, mirror, or tiebreaker) from the current IBM Storage Scale cluster. The removed instance should not be reused afterward.

About this task

You must run this operation from the IBM Storage Scale server instance that continues to exist. This operation removes all IBM Storage Scale server nodes and all disks associated with the removed instance from the cluster.

You can move cluster manager and the primary or secondary server when upgrading an instance. Updating the cluster manager must be run from the command line at the moment because there is no administrator operation that executes the mmchmgr command. For a IBM Storage Scale planned failover and to have a quick IBM Storage Scale cluster recovery, complete these steps.

To avoid the server cluster losing quorum and file system being unavailable temporarily for the IBM Storage Scale clients, ensure the following aspects before running the remove operation:

The IBM Storage Scale server nodes that are part of the instance to be removed should not be part of the quorum.
- Run the Get Cluster Status operation on the server. For more information, see Get Cluster Status operation.
- Verify if the nodes to be removed are marked as quorum. To remove the nodes from quorum run, see Change Quorum State operation.

The IBM Storage Scale server nodes that are part of the instance to be removed should not be primary or secondary servers.
Update the primary and secondary servers to use other nodes using this operation. For more information, see Move Primary and Secondary to Instance.

The IBM Storage Scale server nodes that are part of the instance to be removed should not be cluster or file system manager.
Run these operations to change the cluster and file system manager to other nodes:
- Move Cluster Manager to Instance operation
- Move file system Manager to Instance operation

If the above two commands are not available, ssh into any of the IBM Storage Scale server nodes as root and run:

su - gpfsprod -c 'sudo /usr/lpp/mmfs/bin/mmchmgr -c newIP'
su - gpfsprod -c 'sudo /usr/lpp/mmfs/bin/mmchmgr fileSystemName newIP'

newIP is a server IP that is not part of the instance that is being removed.

You can remove one configuration at a time, keeping in mind the following condition:

If the Remove Member operation is called to remove an instance that is not available (the instance cannot be reached by IP address), do not reactivate that instance. If the instance is reactivated, problems can occur that prevent the IBM Storage Scale cluster from functioning properly.

Procedure

Expand the Failover Active-Active section of the GPFS_Manager operations.
Expand the Remove Member node.
In the Member Type field, select the instance type (Mirror, Tiebreaker, or Passive) to remove from the cluster.
Click Submit.
Click Confirm when asked to validate your request.

Results

A new entry is displayed under the Operation Execution Results list, with the name Remove Member. When completed, the status of the operation changes from Active to Done, and the Return Value column displays the result of the operation.