GPFS cluster manager
There is one GPFS cluster per manager, which is selected from a set of quorum nodes that are designated for the cluster.
For more information, see Quorum.
- Monitors disk leases.
- Detects failures and manages recovery from node failure within the cluster.
The cluster manager determines whether a quorum of nodes exists to allow the GPFS daemon to start and for file system usage to continue.
- Distributes certain configuration changes that must be known to nodes in remote clusters.
- Selects the file system manager node.
The cluster manager prevents multiple nodes from assuming the role of file system manager to avoid data corruption. The token management service resides on the file system manager node and any other nodes that you specify. For more information, see Using multiple token servers.
- Handles UID mapping requests from remote cluster nodes.
- Aggregates health information from all nodes in the cluster.
To identify the cluster manager, issue the mmlsmgr -c command. For more information, see mmlsmgr command.
To change the cluster manager, issue the mmchmgr -c command. For more information, see mmchmgr command.