In some cases, physical restore can fail with the following error:

this happens after PBM starts a node in standalone mode as part of the restore operation, and then tries to shut it down to continue the restore.

The relevant part of the code is

We should perform that shutdown with force: true to make sure it completes. Otherwise cluster is left in broken state.

Refinement comment:

• The “force” usage does not have to be the best solution, we might investigate a more graceful solution

• maybe we should try some retries before the force shutdown

• can we reproduce it? it will be complex anyway…