The v2 worker was not waiting for a resource to be set to PENDING_UPDATE before updating it effectively. In some circumstances (loaded server/slow DB for instance), the resources may have been updated with old data from the DB, resulting in a no-op update. This wait for a PENDING_UPDATE state exists in amphorav1 but it was not moved to v2 when we switched the logic from the DB-based objects to dict-based objects. Story 2009887 Task 44645 Story 2009985 Task 45032 Change-Id: I433e074c2eac5f6875ee5dace8faf86969b24f66
287 B
287 B