We will be documenting about using[ openstack optimize service host maintenance strategy to migrate workload from EDPM compute as a part of post adoption RHEL Minor update task|https://issues.redhat.com/browse/RHOSSTRAT-1137]

Currently host maintenance strategy allows migrating workload from single compute node. Once the workload migration finishes, it disables the maintenance node

For a production deployment, there will be multiple compute nodes. They want to move workloads from multiple compute nodes.

For that we can use zone migration strategy. But after workload migration, it does not disables the nodes. During maintenance phase, New workload might get scheduled on the maintenance node causing unwanted issues during maintenance.

Zone migration strategy already have multiple params for migrating compute and volume nodes. We donot want to add another param here to disable the compute node as it will clutter the interface.

Instead we want to improve host maintenance strategy to accept multiple maintenance node and respective back up node. Let host maintenance strategy do the maintenance and disable the maintenance after workload migration.

Here is what the new interface is going to look:

openstack optimize audit create \
  --strategy host_maintenance \
  --goal cluster_maintaining \
  -p compute_nodes=\
'[{"maintenance_node":"compute1.ctlplane.localdomain", "backup_node":"compute2.ctlplane.localdomain"},
  {"maintenance_node":"compute3.ctlplane.localdomain", "backup_node":"compute4.ctlplane.localdomain"}
  ...
  ...
 ]'

It is not going to deprecate the existing workflow of migrating workload from single compute node.