As discussed in the SciOps workflow introduction meeting.
Currently, dj_worker clears DuplicateErrors under the assumption that we get a DuplicateError from possible job reservation overlaps. This would also, however, clear genuine user-generated DuplicateError.
I propose giving dj_worker a memory of cleared DuplicateErrors and skip those that have been previously cleared.
As discussed in the SciOps workflow introduction meeting.
Currently, dj_worker clears DuplicateErrors under the assumption that we get a DuplicateError from possible job reservation overlaps. This would also, however, clear genuine user-generated DuplicateError.
I propose giving dj_worker a memory of cleared DuplicateErrors and skip those that have been previously cleared.