If SyncJob or BulkJob (and maybe others?) encounter errors, they exit prematurely (leaving stale locks behind) instead of capturing the errors to the Job History.

1. For SyncJob, this is reproducible with a validation error, such as an inbound ou value not being permitted by an AttributeEnumeration configuration.
2. For BulkJob, this is reproducible by trying to expunge the same record twice (the second one will cause an "already deleted" error).
3. Check other CoreJobs for similar behavior.