Rejoin

Rejoin means to become an active member of an IBM® i cluster after having been a nonparticipating member.

For example, when clustering is restarted on a node after the node has been inactive, the cluster node rejoins the cluster. You start cluster resource services on a node by starting it from a node that is already active in the cluster. Beginning with cluster version 3, a node can start itself and will be able to rejoin the current active cluster, provided it can find an active node in the cluster. See Start a cluster node for details.

Suppose that nodes A, B, and C make up a cluster. Node A fails. The active cluster is now nodes B and C. Once the failed node is operational again, it can rejoin the cluster when the node is started from any cluster node, including itself. The rejoin operation is done on a cluster resource group basis, which means that each cluster resource group (CRG) joins the cluster independently.

The primary function of rejoin ensures that the CRG object is replicated on all active recovery domain nodes. The rejoining node, as well as all existing active cluster nodes, must have an identical copy of the CRG object. In addition, they must have an identical copy of some internal data.

When a node fails, the continued calling of cluster resource services on the remaining nodes in the cluster can change the data in a CRG object. The modification must occur due to the calling of an API or a subsequent node failure. For simple clusters, the rejoining node is updated with a copy of the CRG from some node that is currently active in the cluster. However, this may not be true in all cases.

See Starting or ending a cluster administrative domain node for information about the behavior of a cluster administrative domain during a rejoin.