Change Hardware Components for a Controller HostΒΆ

You can replace controller hosts or hardware components while the system is running.

With the exception of StarlingX Simplex, all StarlingX systems require two controllers. You can lock and remove one controller temporarily to replace or upgrade hardware, including primary or secondary disks.

Depending on the type of operation, you may need to delete the host from the Host Inventory. For more information, see Configuration Changes Requiring Re-installation. If you are performing an operation that requires the host to be deleted and re-added, record the current partitioning and volume group assignments for all disks so that you can reproduce them later.

Note

If TPM has been enabled on your StarlingX, and if a controller is replaced or returned to service after a lapse during which a certificate has expired, then TPM must be reconfigured on the StarlingX using the command system certificate-install.

Procedure

  1. Lock the standby controller.

    1. On the Admin menu of the Horizon Web interface, in the System section, select Inventory.

    2. Select the Hosts tab.

    3. In the Actions column, open the drop-down list for the host, and then select Lock Host.

    4. Wait for the host to be reported as Locked.

    The standby controller is shown as Locked, Disabled, and Online.

  2. Power down the host manually and make any required hardware changes.

    StarlingX does not provide controls for powering down a host. Use the BMC or other control unit.

  3. For an operation that affects the Host Inventory record, delete the host from the inventory.

    The Host Inventory contains database information associated with an existing host, such as the MAC address of the management interface NIC, or the presence of StarlingX software on the primary disk. To update this information, you must delete the host using host-delete and then reconfigure and re-add it to the system.

    You must delete the host and then re-add it to the system if a NIC is replaced or moved on a host.

    Note

    Ensure that the host is Online so that its disk is erased when it is deleted from the inventory. This ensures that the host boots from the network when it is powered up for re-installation. If the host is not online when it is deleted from the inventory, you may need to force a network boot during re-installation.

    In the Actions column, open the drop-down list for the host, and select Delete Host.

    The standby controller is removed from the Hosts list, and the StarlingX software is removed from its hard drive.

  4. Reinstall the host.

  5. Power up the host.

    If the host has been deleted from the Host Inventory, the host software is reinstalled.

    Wait for the host to be reported as Locked, Disabled, and Online.

  6. If required, configure the Ceph monitor location.

    Before attempting to unlock the controller, be sure to specify the correct disk for the Ceph monitor, if required.

    Caution

    You must do this before unlocking the reinstalled controller for the first time. Otherwise, the controller reboots continuously on unlock, and must be installed again.

    To specify the correct disk, use a command of the following form:

    ~(keystone_admin)$ system ceph-mon-modify <controller_name> device_node=<diskUUID>
    

    For example:

    ~(keystone_admin)$ system ceph-mon-show controller-1
    +--------------+--------------------------------------+
    | Property     |                                Value |
    +--------------+--------------------------------------+
    | uuid         | ce4a1913-ce1f-4fda-90c0-c49f313d0adc |
    | device_path  | None                                 |
    | device_node  | None                                 |
    | ceph_mon_gib | 30                                   |
    | created_at   | 2016-10-15T00:16:56.423442+00:00     |
    | updated_at   | None                                 |
    +--------------+--------------------------------------+
    ~(keystone_admin)$ system ceph-mon-modify controller-1 device_node=cbc483ad-d7cb-47a8-8622-8846d9444f27
    +--------------+--------------------------------------------+
    | Property     | Value                                      |
    +--------------+--------------------------------------------+
    | uuid         | ce4a1913-ce1f-4fda-90c0-c49f313d0adc       |
    | device_path  | /dev/disk/by-path/pci-0000:00:0d.0-ata-3.0 |
    | device_node  | /dev/sdc                                   |
    | ceph_mon_gib | 30                                         |
    | created_at   | 2016-10-15T00:16:56.423442+00:00           |
    | updated_at   | None                                       |
    +--------------+--------------------------------------------+
    
    System configuration has changed.
    please follow the administrator guide to complete configuring system.
    ~(keystone_admin)$ system ceph-mon-show controller-1
    
    +--------------+--------------------------------------------+
    | Property     | Value                                      |
    +--------------+--------------------------------------------+
    | uuid         | ce4a1913-ce1f-4fda-90c0-c49f313d0adc       |
    | device_path  | /dev/disk/by-path/pci-0000:00:0d.0-ata-3.0 |
    | device_node  | /dev/sdc                                   |
    | ceph_mon_gib | 30                                         |
    | created_at   | 2016-10-15T00:16:56.423442+00:00           |
    | updated_at   | 2016-10-15T00:35:44.181413+00:00           |
    +--------------+--------------------------------------------+
    
  7. Unlock the host to make it available for use.

    On the Hosts tab of the Host Inventory page, open the drop-down list for the host, and then select Unlock Host.

    The host is rebooted, and its Availability State is reported as In-Test. After a few minutes, it is reported as Unlocked, Enabled, and Available.

  8. If the same hardware change is required on both controllers, make the change to the other controller.

    1. Open the drop-down menu for the active controller and then select Swact Host.

      Up to 20 minutes can be required to complete the swact.

      Note

      During the swact, access to Horizon is temporarily interrupted, and the login screen is displayed. Wait for a few minutes, and then log in. The new active controller is shown as Degraded, and then changed to Available.

      The Controller-Active and Controller-Standby personalities are updated in the Hosts List.

    2. Return to Step 1 and repeat the procedure for the new standby controller.

Results

The updated controllers are now in service.