Rehome a Subcloud¶
When the System Controller needs to be reinstalled, or when the subclouds from multiple System Controllers are being consolidated into a single System Controller, you can add already deployed subclouds to a different System Controller using the rehoming playbook.
Note
The rehoming playbook does not work with freshly installed/bootstrapped subclouds.
Note
The system time should be accurately configured on the System Controllers and the subcloud’s controllers before rehoming the subcloud.
Note
The system controllers and subclouds must run the same software versions.
Warning
Do not rehome a subcloud if the RECONCILED status on the system resource or any host resource of the subcloud is FALSE. To check the RECONCILED status, run the kubectl -n deployment get system and kubectl -n deployment get hosts commands.
Use the following procedure to enable subcloud rehoming and to update the new subcloud configuration (networking parameters, passwords, etc.) to be compatible with the new System Controller.
About this task
There are six phases for Rehoming a subcloud:
Unmanage the subcloud from the previous System Controller.
Note
You can skip this step if the previous System Controller is no longer running or is unable to connect to the subcloud.
Update the admin password on the subcloud to match the new System Controller, if required.
Run the subcloud add command with the
--migrate
option on the new System Controller. This will update the System Controller and connect to the subcloud to update the appropriate configuration parameters.Use the dcmanager subcloud list command to check the status of the subcloud, ensure the subcloud is online and complete before managing the subcloud.
Delete the subcloud from the previous System Controller after the subcloud is offline.
Note
You can skip this phase if the previous System Controller is no longer running or is unable to connect to the subcloud.
On the new System Controller, set the subcloud to “managed” and wait for it to sync.
Prerequisites
Ensure that the subcloud management subnet, oam_floating_address, oam_node_0_address and oam_node_1_address (if applicable) does not overlap addresses already being used by the new System Controller or any of its subclouds.
Ensure that the subcloud has been backed up, in case something goes wrong and a subcloud system recovery is required.
Ensure that the system time between new system controllers and the subclouds are accurately configured.
~(keystone_admin)]$ date -u
If the time is not correct either on the system controllers or the subclouds, check the system’s
clock_synchronization
config on the system.~(keystone_admin)]$ system host-show controller-0
Check the NTP server configuration or PTP server configuration sections to correct the system time based on the system’s
clock_synchronization
config (NTP or PTP).Transfer the yaml file that was used to bootstrap the subcloud prior to rehoming, to the new System Controller. This data is required for rehoming.
If the subcloud can be remotely installed via Redfish Virtual Media service, transfer the yaml file that contains the install data for this subcloud, and use this install data in the new System Controller, via the
--install-values
option, when running the remote subcloud reinstall, upgrade or restore commands.
Note
These prerequisites apply if the old System Controller is still available.
Procedure
If the previous System Controller is running, use the following command to ensure that it does not try to change subcloud configuration while you are modifying it to be compatible with the new System Controller.
~(keystone_admin)]$ dcmanager subcloud unmanage <subcloud_name>
Ensure that the subcloud’s bootstrap values file is available on the new System Controller. If required, in the subcloud’s bootstrap values file update the
systemcontroller_gateway_address
entry to point to the appropriate network gateway for the new System Controller to communicate with the subcloud.If the admin password of the subcloud does not match the admin password of the new System Controller, use the following command to change the subcloud admin password. This step is done on the subcloud that is being migrated.
~(keystone_admin)]$ openstack user password set
Note
You will need to specify the old and the new password.
For an AIO-DX subcloud, ensure that the active controller is controller-0. Perform a host-swact of the active controller (controller-1) to make controller-0 active.
~(keystone_admin)]$ system host-swact controller-1
Ensure that all the subcloud controllers are online and available by the system host-list command, and free of “250.001 config out-of-date” alarm by the fm alarm-list command. If there’s “250.001 config out-of-date” alarm, a lock/unlock is required to clear that alarm on the node before the next step.
On the new System Controller, use the following command to start the rehoming process.
~(keystone_admin)]$ dcmanager subcloud add --migrate --bootstrap-address <subcloud-controller-0-oam-address> --bootstrap-values <bootstrap_values_file> [--install-values <install_values_file>]
Note
You will need to update the
systemcontroller_gateway_address
variable in the bootstrap values file before you perform the migration. This field is the gateway address to the new System Controller.The subcloud deploy status will change to “pre-rehome” and if the preliminary steps complete successfully it will change to “rehoming”. At this point an Ansible playbook will run and update the appropriate configuration data in the subcloud. You can query the status by running dcmanager subcloud show command. Once the subcloud has been updated, the subcloud deploy status will change to “complete”.
Note
The
--install-values
parameter is optional and is not mandatory for subcloud rehoming. However, you can opt to save these values on the new System Controller as part of the rehoming process so that future operations that involve remote reinstallation of the subcloud (e.g. reinstall, upgrade, restore) can be performed for the rehomed subcloud.The subcloud install values can also be added to or updated on the new System Controller using the dcmanager subcloud update --install-values command post subcloud rehoming.
Delete the “image:” line from the install-values file, if it exists, so that the image is correctly located based on the new System Controller configuration.
If the previous System Controller is still running, delete the subcloud after it goes offline, using the following command.
~(keystone_admin)]$ dcmanager subcloud delete <subcloud-name>
Use the dcmanager subcloud list command to display the status of the subcloud, and ensure the subcloud is online and complete before managing the subcloud.
~(keystone_admin)]$ dcmanager subcloud list +----+-----------+------------+--------------+---------------+---------+ | id | name | management | availability | deploy status | sync | +----+-----------+------------+--------------+---------------+---------+ | 1 | subcloud1 | unmanaged | online | complete | unknown | +----+-----------+------------+--------------+---------------+---------+
Use the following command to “manage” the subcloud. This is executed on the System Controller.
~(keystone_admin)]$ dcmanager subcloud manage <subcloud-name>
The new System Controller will audit the subcloud and determine whether it is in-sync with the System Controller.
Error Recovery
If the subcloud rehoming process begins successfully, (status changes to “rehoming”) but there is a transient fault that prevents step 5 from completing successfully, then manual error recovery is required.
The first stage of error recovery is to delete the subcloud from the new System Controller and re-attempt rehoming using the following commands:
~(keystone_admin)]$ dcmanager subcloud delete <subcloud-name>
~(keystone_admin)]$ dcmanager subcloud add --migrate --bootstrap-address <subcloud-controller-0-oam-address> --bootstrap-values <bootstrap_values_file> [--install-values <install_values_file>]
If all attempts fail, restore the subcloud from backups, that will revert the subcloud to the original state prior to rehoming.