Enqueuer job: improve aborted migrations handling
🔥 Problem
The enqueuer will pick aborted migrations to retry them.
The challenge is that the backend need to choose which step to retry: pre_import
or import
. To take this decision, the analysis described a way: look at the recorded migration timestamps of the container image.
The problem is that those timestamps could be unsynced with what happend on the container registry side due to restarts, request timeouts or other.
Thus, those timestamps should be considered as unreliable.
🚒 Solution
Use this endpoint to get the real status of the migration and take the proper action out of the result of that call.