Clearing preexisting split-brain condition

Review the information on how the VxFEN driver checks for preexisting split-brain condition.

See Fencing startup reports preexisting split-brain.

Table: Recommened solution to clear pre-existing split-brain condition describes how to resolve a preexisting split-brain condition depending on the scenario you have encountered:

Table: Recommened solution to clear pre-existing split-brain condition

Scenario

Solution

Actual potential split-brain condition - system 2 is up and system 1 is ejected

  1. Determine if system1 is up or not.

  2. If system 1 is up and running, shut it down and repair the private network links to remove the split-brain condition.

  3. Restart system 1.

Apparent potential split-brain condition - system 2 is down and system 1 is ejected

(Disk-based fencing is configured)

  1. Physically verify that system 2 is down.

    Verify the systems currently registered with the coordination points.

    Use the following command for coordinator disks:

    # vxfenadm -s all -f /etc/vxfentab

    The output of this command identifies the keys registered with the coordinator disks.

  2. Clear the keys on the coordinator disks as well as the data disks in all shared disk groups using the vxfenclearpre command. The command removes SCSI-3 registrations and reservations.

    See About the vxfenclearpre utility.

  3. Make any necessary repairs to system 2.

  4. Restart system 2.

Apparent potential split-brain condition - system 2 is down and system 1 is ejected

(Server-based fencing is configured)

  1. Physically verify that system 2 is down.

    Verify the systems currently registered with the coordination points.

    Use the following command for CP servers:

    # cpsadm -s cp_server -a list_membership 
    -c cluster_name

    where cp_server is the virtual IP address or virtual hostname on which CP server is configured, and cluster_name is the VCS name for the VCS cluster (application cluster).

    The command lists the systems registered with the CP server.

  2. Clear the keys on the CP servers using the cpsadm command. The cpsadm command clears a registration on a CP server:

    # cpsadm -s cp_server -a unreg_node 
    -c cluster_name -n nodeid

    where cp_server is the virtual IP address or virtual hostname on which the CP server is listening, cluster_name is the VCS name for the VCS cluster, and nodeid specifies the node id of VCS cluster node. Ensure that fencing is not already running on a node before clearing its registration on the CP server.

    After removing all stale registrations, the joiner node will be able to join the cluster.

  3. Make any necessary repairs to system 2.

  4. Restart system 2.