Description

During a Zero Impact Upgrade (ZIU), the process stalled at Stage 1. The upgrade was unable to progress as the contrail-controller did not receive the expected stage completion signal from one of the contrail-agent units.

Symptoms

 

  • In a Juju-Based Lifecycle Orchestration setup, each Contrail component (e.g., contrail-agent, contrail-analytics, contrail-controller) is managed via Juju charms.
  • During a ZIU, each unit must report its upgrade progress (e.g., ziu_stage_set(1)) to the Juju controller. The Juju controller then communicates this to the contrail-controller charm to coordinate the upgrade.
  • In this case, unit contrail-agent-* ( failed to send or complete its stage update. (Please check unit-contrail-agent* logs to check the behaviour)
  • Log snippet

xx-xx-xx INFO unit.contrail-controller/xxx.juju-log server.go:316 ZIU: stage is not ready: rel=contrail-controller:xxx unit=contrail-agent/x value=0

The contrail-controller waited indefinitely for this signal, causing the upgrade process to stall at Stage 1.

 

Solution

 

  • Restart the Juju controller, which restored communication and allowed the ZIU to proceed normally.
  • The upgrade completed successfully

 

Modification History

2025-08-12 : Article Created