Removing a HA controller node from a deployed topology

After deploying your high availability (HA) controller +n compute topology, you can remove a HA controller node system from your deployment. This only applies to the high availability (HA) controller +n compute topology.

About this task

The knife os manage remove node command attempts to stop and disable any cluster services and running OpenStack services on the HA controller node. Then, it removes the HA controller node from the cloud topology, and also removes the chef 'node' and 'client' objects for the node from the Chef server.
After the HA controller node is removed, the command attempts to update the topology to finish the process of removing the HA controller node from the cluster. The update process attempts to refresh the entire topology that is defined in the topology JSON file. If any HA controller nodes in the topology are offline or are currently powered off, the update process fails unless it is instructed to not update those nodes. To instruct the update process to not update a node, you can set a key in the topology JSON file for that particular node to specify whether to update it.
Node Keys: allow_update
A Boolean value, which indicates whether to update the node. Set to true to update the node. Set to false if you do not want to update the node. The default is true.
Note: If the topology file must be modified to not update particular nodes in the topology, that change must be complete before attempting to remove the HA controller node.

For more information about the topology update process, see Updating a deployed topology. For more information about the topology JSON file format, see Topology JSON file.

Note: Removing an HA controller node with IBM DB2® is not supported. To determine if an HA controller node includes IBM DB2, run the knife node show node-fqdn command on the deployment server. Change node-fqdn to the fully qualified domain name of the node system in your topology. If the node tags include ibm-os-deploy-ha-db, then the node includes IBM DB2.

Use the following procedure to remove an HA controller node to your deployment.

Procedure

  1. Log in to the deployment system as the root user.
    This is the system where IBM Cloud Manager with OpenStack was installed.
  2. Remove the HA controller node.
    Change the following command parameters.
    • topology-file: Set to the topology file used for deployment. Specifying the topology-file parameter is required when removing a HA controller node. The command removes the HA controller node information from the topology file. 
      knife os manage remove node node-fqdn --topology-file topology-file

Results

The HA controller node is now removed from the deployed topology environment. To clean up the removed HA controller node, see Cleaning up a node for redeployment.