Node class management

Each mmvdisk recovery group must be associated with an mmvdisk node class that contains only the servers for the recovery group.

In the case of paired recovery groups, each recovery group in the pair will be associated with the same node class since it contains the primary and backup servers for that paired recovery group.

The mmvdisk nodeclass create command is used to create an mmvdisk node class that contains the intended servers for a recovery group or recovery group pair.

The node class is used by the mmvdisk server configure command to configure IBM Storage Scale RAID on the recovery group servers.

The association between an mmvdisk node class and an mmvdisk recovery group can be established in two ways:
  1. The mmvdisk nodeclass create command is used to create a node class containing the intended servers for the recovery group, and the mmvdisk recoverygroup create command is used to create an mmvdisk recovery group using the node class.
  2. The mmvdisk recoverygroup convert command is used to convert two paired recovery groups and their two servers to mmvdisk management. The mmvdisk recoverygroup convert command must be supplied with a node class name to be used as the mmvdisk node class for the converted recovery group pair.
Once a recovery group is created using an mmvdisk node class, the members of the node class can only be changed using the mmvdisk recoverygroup command.

When mmvdisk file systems span multiple recovery groups or recovery group pairs, each recovery group or recovery group pair is a single point of failure for its vdisk NSDs. In the case of a shared or scale-out recovery group, it is simply that the recovery group must be available for its vdisk NSDs to be available. In the case of a recovery group pair, the disk enclosures twin-tailed between the two servers are the single point of failure since each enclosure is split between the two recovery groups and will hold vdisk NSD data for both of the paired recovery groups. A shared recovery group can also be characterized as a single point of failure because of its twin-tailed enclosure.

The mmvdisk node class captures this relationship perfectly and is used by the mmvdisk filesystem command to automatically assign the failure groups for vdisk NSDs when a file system spans multiple recovery groups or recovery group pairs. Each vdisk NSD from the same node class is assigned the same file system failure group. If custom failure group numbers are desired, the mmvdisk filesystem command permits them to be assigned using the mmvdisk node class names as identifiers for the single point of failure.

The mmvdisk node class behaves in all other respects like a regular IBM Storage Scale node class.