This bug happens sometimes but was recently reproducible in CI, using a continuous audit with a small interval (5 or 10 seconds). The strategy is not properly treating a exception raised by the model and with that, the audit goes to a FAILED state, which invalidates the continuous audit.

To Reproduce Steps to reproduce the behavior:

1. Create a continuous audit, with interval of 10 seconds, setting the goal to "hardware_maintenance" and strategy to "zone_migration".
2. Create a new instance (or volume) in the source host. Maybe more instances are needed until it reproduce the issue.
3. The issue happens when the strategy is able to list the nova instance but it is not yet available in the model, raising an exception in decision-engine logs [1]
4. The audit goes to FAILED state.

Expected behavior

• While the instance is not available in the model, the strategy will not consider it as a possible migration from the source host.

Bug impact

• The continuous audit will go to a FAILED state, and stop proposing solutions.

Known workaround

• None

Additional context

• Upstream LP Bug: https://bugs.launchpad.net/watcher/+bug/2098984

[1] https://bugs.launchpad.net/watcher/+bug/2098984