Unable to Initiate Compute Driver: "Node <uuid> could not be found"

Problem

  • Nova instances in the baremetal region are stuck in "BUILD" state.

  • The baremetal controller node is observed to be in a failed state in the Platform9 dashboard and/or via Resource Manager (Resmgr) API.

  • /var/log/pf9/hostagent.log shows "In failed state until next set_config message...". Prior to this, the pf9-ostackhost and pf9-novncproxy services may show as continuously setting their service state, e.g.

2023-08-04 02:18:58,255 - pf9_app.py INFO - Setting the desired service state
2023-08-04 02:18:58,256 - pf9_app.py INFO - Setting service state pf9-ostackhost.5.6.6-4803. Command: sudo systemctl start pf9-ostackhost
2023-08-04 02:18:58,422 - pf9_app.py INFO - Setting service state pf9-novncproxy.5.6.6-4803. Command: sudo systemctl stop pf9-novncproxy
2023-08-04 02:19:05,361 - session.py INFO - Converge succeeded
2023-08-04 02:20:14,895 - session.py INFO - In failed state until next set_config message...
  • Meanwhile, the /var/log/pf9/ostackhost.log is logging an error similar to the following.

INFO nova.service [-] Starting compute node (version 5.6.0)
ERROR oslo_service.service [req-8b0516ca-a439-4446-b529-8f4ebbad2b4e None None] Error starting thread.: openstack.exceptions.ResourceNotFound: No Node found for 8af46697-7602-40c4-8350-eb48423e9822: Client Error for url: http://localhost:8158/ironic/v1/nodes/8af46697-7602-40c4-8350-eb48423e9822?fields=uuid%2Cpower_state%2Ctarget_power_state%2Cprovision_state%2Ctarget_provision_state%2Clast_error%2Cmaintenance%2Cproperties%2Cinstance_uuid%2Ctraits%2Cresource_class, Node 8af46697-7602-40c4-8350-eb48423e9822 could not be found.

Environment

  • Platform9 Managed OpenStack - v5.6.6

  • Ironic/Baremetal Service

  • Nova

Cause

If the baremetal node was re-provisioned, and the associated Nova instance was not deleted/cleaned up, then, this error is to be expected.

If the baremetal node was not re-provisioned at any point, please contact Platform9 Support for further analysis.

Resolution

Baremetal Node was Re-Provisioned

  1. Delete the associated Nova instance (via the Platform9 UI, or OpenStack CLI).

  2. Restart the pf9-ostackhost service on the baremetal controller node.