Shutting Down a Clustered Configuration (CLI)

Use this procedure to shut down a clustered configuration. A cluster that has been shut down can be restored to the same cluster configuration as before the shutdown by powering on both controllers.

  1. Go to configuration cluster.
  2. Determine the cluster state of each controller.

    In the following example, controller-a is the owner and in the active state. Its peer, controller-b, is the standby controller and in the stripped state.

    controller-a:> configuration cluster
    controller-a:configuration cluster> get
                             state = AKCS_OWNER
                       description = Active (takeover completed)
                          peer_asn = 365ed33c-3b9d-c533-9349-8014e9da0408
                     peer_hostname = controller-b
                        peer_state = AKCS_STRIPPED
                  peer_description = Ready (waiting for failback)

    The following table describes the pairs of controller states (state and peer_state) that you could see.

    controller-a controller-b Condition

    AKCS_CLUSTERED

    AKCS_CLUSTERED

    Both controllers are running in a normal clustered condition.

    AKCS_OWNER

    AKCS_STRIPPED

    controller-a owns all of the resources and is the active controller.

    controller-b is in standby mode and has no resources. To limit the number of times a pool is moved, shut down the STRIPPED controller first.

    AKCS_OWNER

    REBOOTING

    controller-a has all resources.

    controller-b is rebooting.

    AKCS_OWNER

    UNKNOWN

    controller-a has all resources.

    controller-b is powered off or rebooting, or all of its cluster interconnect links are down.

  3. Shut down controller-b.
    controller-b:configuration cluster> cd /
    controller-b:> maintenance system poweroff
    This will turn off power to the appliance. Are you sure? (Y/N) Y

    Note:

    If both controllers have a status of AKCS_CLUSTERED, a takeover of the surviving controller begins automatically.
  4. From controller-a, verify that controller-b has been powered off and is in state OWNER/unknown.
    controller-a:configuration cluster> get
                             state = AKCS_OWNER
                       description = Active (takeover completed)
                          peer_asn = 365ed33c-3b9d-c533-9349-8014e9da0408
                     peer_hostname = controller-b
                        peer_state = OWNER/unknown
                  peer_description = 
  5. Shut down controller-a.
    controller-a:configuration cluster> cd /
    controller-a:> maintenance system poweroff
    This will turn off power to the appliance. Are you sure? (Y/N) Y
  6. Optional: Use Oracle ILOM to confirm that both controllers are powered off.

    Enter the following command in the Oracle ILOM CLI:

    ->show /SYS power_state

    For information about accessing Oracle ILOM, see Logging in to Oracle ILOM Remotely Using a Command Line Interface in Oracle ZFS Storage Appliance Customer Service Manual, Release OS8.8.x.

  7. Power off the disk shelves.
    1. Place the power supply on/off switches to the "O" off position.
    2. Disconnect the power cords from the external power source for the cabinet.

      Note:

      All power cords must be disconnected to completely remove power from the disk shelf.

      For more information, see Powering Off a Disk Shelf in Oracle ZFS Storage Appliance Installation Guide, Release OS8.8.x.