Personal data ends up in every derived store, and only the primary database is covered by a delete statement.
the stores a deletion has to reach:
the database covered by the obvious query
the search index a separate delete, by id
the cache a key pattern, or a version bump
the queue a job payload holding the name
the logs a retention window, and no delete
the error tracker an API call, per event
the analytics platform their API, and their timeline
the backups no answer, only a retention window
The queue is the one that surprises people: a job enqueued before the request contains the personal data in its payload and will run afterwards. Logs and backups have no clean answer and the honest position is a documented retention window rather than a claim of deletion — which is what most regulators expect and what most privacy policies do not say.