In the event a worker suddenly fails, it will leave the task it was running with state "IN PROGRESS" in DB.
Such tasks must be automatically reset to "PENDING" state to be processed by another worker by a job which may run every 10 minutes.
This job will rely on the UUID of the workers which are currently running to reset tasks of any other worker UUID:
- when running a standalone SonarQube, the UUIDs for the workers is available in memory.
- when running a SonarQube cluster, the UUIDs of the workers must be shared across all CE nodes so that any of them is capable of performing this reset responsibility