- Large number of small objects submitted as a Job Manifest -- This will test ZK interactions with multiple processes and locks
- Small number of very large Jobs -- This will test for timeout conditions
- Errors in Batch creation should fail all jobs and notify
- Jobs set to fail are represented correctly in ZK batch and queue nodes
- Ensure that failed jobs release all ZK locks
- Test hold collection before starting batch
- Test hold collection as a large batch is processing
- Release job while collection is held
- Release job after collection hold is released
- Release all jobs for a collection
- Requeue job failure for each phase (document how to force failures)
- Estimating (temp hardcode)
- Provisioning (temp hardcode)
- Downloading (bad manifest url)
- Processing (bad storage node)
- Recording (ask David, turn off inventory and manually fail the step)
- Notify (possibly bad email or callback)