Description
Context:
During random upgrade test failure I encountered a bug that can happen when the status endpoint is called when concurrently the import is still initialized
See attached logs that also contains explanations.
AT:
- EngineImportSchedulerFactory.getImportSchedulers should either by synchronized or even be part of the initialization of the bean (or both) so we can't run into such race-condition
Hint:
This can get caused either by container health checks using that endpoint (like it did for me during testing) but also by userRequests when Optimize starts (which may happen on restarts).