1262938 - reference_data_signature table should have a unique index on ``signature``

Status: NEW

(Tree Management :: Treeherder: Data Ingestion, defect, P3)

Description

[Cameron Dawson [:camd]]

Parts of our code make the assumption that the job signature is unique for a job with certain attributes.  But that hasn't always been the case.  Bug 1242038 makes this true going forward.  But until that's pushed to production, we will continue to have duplicate signatures in that table.

Once that's been pushed out, we should scrub that table and delete all but the latest signature entry for each signature so that the field is unique.  Then we should create a unique index on that field to ensure it stays that way going forward.

Comment 1

[William Lachance (:wlach)]

When this is done, it would be nice to turn this statement from a filter to a get:

https://github.com/mozilla/treeherder/blob/464c82754ed947e7f5f4216027f2eb121ddd2e19/treeherder/model/derived/artifacts.py#L118