respect last_run_at on save

[diaxoaine]

Updating an existing entry that has already run recalculates the schedule score with the last_run_at = now(), while still keeping the old last_run_at in the meta. The entry ends up waiting a full schedule cycle before firing again.

Current behaviour:

I have this hourly task

from redbeat import RedBeatSchedulerEntry
interval = celery.schedules.schedule(run_every=3600)
entry = RedBeatSchedulerEntry('task-name', 'tasks.some_task', interval, args=['args'])
entry.save()

The last run was 45 mins ago, so it will run in 15 mins from now. And now I update its args:

entry = RedBeatSchedulerEntry('task-name', 'tasks.some_task', interval, args=['new'])
entry.save()

the score is now 1 hour from now, instead of 15 mins.

Expected behaviour:

The score should still be 15 mins from now, since only the args changed.

[sibson]

Your creating a new task, not updating the existing one. Use entry = RedBeatSchedulerEntry.from_key(), to update an existing entry

[diaxoaine]

Your creating a new task, not updating the existing one. Use entry = RedBeatSchedulerEntry.from_key(), to update an existing entry

I'd say RedBeatSchedulerEntry(...) + save() already behaves like an upsert because entry's name is unique.