Deleting nodes from a GPFS cluster

You can delete nodes from a GPFS cluster by issuing the mmdelnode command.

The GPFS daemon must be shut down on a node before the node can be deleted. The following types of nodes cannot be deleted unless some reconfiguration is done:
  • A node that is an NSD server cannot be deleted if it is the only NSD server for one or more NSDs in the cluster. Issue the mmlsnsd command to list the NSD servers and NSDs in the cluster. If a node is the only NSD server for some NSDs, you can issue the mmchnsd command to assign other NSD servers to those NSDs.
  • A node that is a primary or secondary cluster configuration server cannot be deleted. Issue the mmlscluster command to list the primary and secondary cluster configuration servers, if any are configured. If a node is a primary or secondary configuration server, you can issue the mmchcluster command to create a new primary or secondary configuration server.
  • If the GPFS state is unknown and the node is reachable on the network.
    You cannot delete a node if both of the following are true:
    • The node responds to a TCP/IP ping command from another node.
    • The status of the node shows unknown when you use the mmgetstate command from another node in the cluster.
    Note: You can delete such a node if you physically power it off.
  • If the node is configured as a performance monitoring collector. In such cases, you need to remove the node from the performance monitoring configuration by using the mmperfmon config update --collectors command before deleting the node. Deleting a collector node causes loss of all the collected perfmon stats data on the collector node.
  • If the node is defined as a Transparent cloud tiering node. You can determine whether a node is a Transparent cloud tiering node by issuing the mmcloudgateway node list command. If the node is listed as the Transparent cloud tiering node, and you still want to delete it without deleting the cluster, first use the mmchnode command to disable the Transparent cloud tiering node role:
    • If the node is a Transparent cloud tiering node, disable Transparent cloud tiering from the node by using the mmchnode --cloud-gateway-disable command, and then uninstall the Transparent cloud tiering rpms. Doing so ensures that the mmdelnode command does not fail on a Transparent cloud tiering node.
  1. To delete the nodes listed in a file called nodes_to_delete, issue the following command:
    mmdelnode -N /tmp/nodes_to_delete
    where nodes_to_delete contains the nodes k164n01 and k164n02. The system displays information similar to the following:
    Verifying GPFS is stopped on all affected nodes ...
    mmdelnode: Command successfully completed
    mmdelnode: 6027-1371 Propagating the cluster configuration data to all
      affected nodes.  This is an asynchronous process.
    
  2. To confirm the deletion of the nodes, issue the following command:
    mmlscluster
    The system displays information similar to following:
    GPFS cluster information
    ========================
      GPFS cluster name:         cluster1.kgn.ibm.com
      GPFS cluster id:           15529849231188177215
      GPFS UID domain:           cluster1.kgn.ibm.com
      Remote shell command:      /usr/bin/ssh
      Remote file copy command:  /usr/bin/scp
      Repository type:           CCR
    
     Node  Daemon node name        IP address     Admin node name         Designation
    -------------------------------------------------------------------------------------
       1   k164n03.kgn.ibm.com   198.117.68.68    k164n03.kgn.ibm.com     quorum
       2   k164n04.kgn.ibm.com   198.117.68.69    k164n04.kgn.ibm.com     quorum
       3   k164n05.kgn.ibm.com   198.117.68.70    k164n05.kgn.ibm.com     quorum-manager
    
  3. If you disabled file audit logging in step 1, you can enable it by following the instructions in Enabling file audit logging on a file system.

For information about deleting protocol nodes (CES nodes) from a cluster, see Deleting a Cluster Export Services node from an IBM Storage Scale cluster.

For more information, see mmdelnode command and mmlscluster command.

Exercise caution when shutting down GPFS on quorum nodes or deleting quorum nodes from the GPFS cluster. If the number of remaining quorum nodes falls below the requirement for a quorum, then you are unable to perform file system operations. For more information about quorum, see Quorum.