We let users delete PII before it gets hashed. Currently, those queued jobs still run (although they exit early) but maybe they shouldn’t.
This sounds trivial, but it’s actually not. Firstly, we need to get access to the tasks to cancel them. We can either store the task ID in the database, adding extra complexity and undesirable redundancy; or we can look for the corresponding task in Celery, which is potentially slow.
Secondly, I’m not sure how efficient Celery is at cancelling tasks.
Aha! Link: https://csiro.aha.io/features/ANONLINK-89
We let users delete PII before it gets hashed. Currently, those queued jobs still run (although they exit early) but maybe they shouldn’t.
This sounds trivial, but it’s actually not. Firstly, we need to get access to the tasks to cancel them. We can either store the task ID in the database, adding extra complexity and undesirable redundancy; or we can look for the corresponding task in Celery, which is potentially slow.
Secondly, I’m not sure how efficient Celery is at cancelling tasks.
Aha! Link: https://csiro.aha.io/features/ANONLINK-89