Create a new maintenance session. You can specify a list of ‘hosts’ to be maintained or have an empty list to indicate those should be self-discovered. You need to have an initial state for the workflow in ‘state’. ‘workflow’ indicates the name of a Python plug-in to be used in the maintenance.

{
    "hosts": ["overcloud-novacompute-1.opnfvlf.org",
              "overcloud-novacompute-2.opnfvlf.org",
              "overcloud-novacompute-3.opnfvlf.org"],
    "state": "MAINTENANCE",
    "maintenance_at": "2018-02-28 06:06:03",
    "metadata": {"openstack_release": "Queens"},
    "workflow": "default",
    "download": ["https://my.sw.upgrades.com/SW1.tar.gz",
                 "https://my.sw.upgrades.com/ESW1.tar.gz",
                 "https://my.sw.upgrades.com/plugins1.tar.gz"],
    "actions": [
        {"plugin": "prepare", "type": "pre"},
        {"plugin": "esw_upgrade", "type": "host", "metadata": {"upgrade": "ESW1"}},
        {"plugin": "os_upgrade", "type": "host", "metadata": {"upgrade": "SW1"}},
        {"plugin": "finalize", "type": "post"}]
}

{
    "session_id": "695030ee-1c4d-11e8-a9b0-0242ac110002"
}

Update existing maintenance session. This can be used to continue a failed session.

Get all ongoing maintenance sessions.

{
    "session_id": ["695030ee-1c4d-11e8-a9b0-0242ac110002"]
}

Get a maintenance session state.

{
    "state": "MAINTENANCE_DONE"
}

Delete a maintenance session. Usually called after the session is successfully finished.

Get project instances belonging to the current state of maintenance session. the Project-manager receives an AODH event alarm telling about different maintenance states. Event data field length is very limited, so instances cannot be given as a list in the event. Instead, there will be an URL given to below API to get a project-specific list of instances.

{
    "instance_ids": ["109e14d9-6566-42b3-93e4-76605f264d8f",
                     "71285107-f0fc-4428-a8b2-0b3edd64bcad"]
}

Project having instances on top of the infrastructure handled by a maintenance session might need to make own action for its instances on top of a host going into maintenance next, or reply an admin action to be done. This is, as the host can go down or even be removed and the instances should be then running safely somewhere else. Project manager receives an AODH event alarm telling which instances are affected and when the project is ready, it makes its own action or replies back an action which needs the admin privileges.

{
    "instance_actions": {"109e14d9-6566-42b3-93e4-76605f264d8f": "MIGRATE",
                         "71285107-f0fc-4428-a8b2-0b3edd64bcad": "MIGRATE"},
    "state": "ACK_PLANNED_MAINTENANCE"
}