Stopping and Starting the HyperScale X Cluster

You can stop and restart a Commvault HyperScale X cluster when required.

Caution

To maintain data integrity and prevent data loss, always follow the shutdown and startup procedures described in this topic when stopping or starting the cluster.

  • If a single node must be stopped for maintenance, use the procedure in Stopping and Starting a Node.

  • If more than one node must be stopped, shut down the entire cluster using this procedure.

  • Improperly shutting down multiple nodes individually can cause data integrity issues, data loss, and restore failures.

Procedure

Perform an operation based on your requirement.

Stopping the Cluster

  1. Stop or suspend all running jobs associated with the MediaAgents in the cluster. For details, see Controlling Jobs.

  2. Set the MediaAgents to maintenance mode. For details, see Setting the MediaAgent on Maintenance Mode.

  3. Stop the Commvault services and verify that the Commvault File System (CVFS) vdisk is unmounted on all nodes:

    1. Using an SSH client (for example, PuTTY), log in to a node with root credentials.

    2. Identify the <vdiskname> using the following command:

      <vdiskname> appears as a subfolder under /ws/hedvig.

      # df -h

      Example output:

      127.0.0.1:/exports/CVLTBackupSMC43002153 269T -1.5T 271T 0% /ws/hedvig/CVLTBackupSMC43002153 127.0.0.1:/exports/CVLTBackupSMC43002153 269T -1.5T 271T 0% /ws/hedvig/CVLTBackupSMC43002153-r

    3. Stop the services:

      # commvault stop
    4. Verify that the vdisk is unmounted by running the mount command. The output should not include <vdiskname>.

    5. Repeat these steps for all nodes in the cluster.

  4. Start hv_deploy mode:

    Note

    On HyperScale 2.x installations, switch to the admin user before stopping nodes: # su - admin.

    # export HV_PUBKEY=1
    # hv_deploy
  5. Display and note the cluster name:

    # show_all_clusters

    The cluster name appears as shown in the following sample output:

    Cluster name: HV13102021054000 - owner: unowned - version: v-4.5.0.0.2.3597.00320e0ba9:06aca75bc4db06170a13ef14cbfe8446

  6. Log in to the cluster:

    # login_to_cluster cluster name

    Example:

    # login_to_cluster HV13102021054000

    When prompted, enter the root password.

  7. Stop the cluster:

    # stop_cluster
  8. Exit hv_deploy mode:

    # exit

Starting the Cluster

  1. Using an SSH client (for example, PuTTY), log in to a node with root credentials.

  2. Start hv_deploy mode:

    Note

    On HyperScale 2.x installations, switch to the admin user before starting nodes: # su - admin.

    # export HV_PUBKEY=1
    # hv_deploy
  3. Display and note the cluster name:

    # show_all_clusters

    The cluster name appears as shown in the following sample output:

    Cluster name: HV13102021054000 - owner: unowned - version: v-4.5.0.0.2.3597.00320e0ba9:06aca75bc4db06170a13ef14cbfe8446

  4. Log in to the cluster:

    # login_to_cluster cluster name

    Example:

    # login_to_cluster HV13102021054000

    When prompted, enter the root password.

  5. Start the cluster:

    # start_cluster
  6. Exit hv_deploy mode:

    # exit
  7. Start the Commvault services (this will also mount the CVFS vdisk.)

    # commvault start
  8. Disable maintenance mode for the MediaAgent to bring it online. For details, see Setting the MediaAgent on Maintenance Mode.

Handling Improper Shutdowns of HyperScale Nodes or Clusters

×

Loading...