When ETL is found to be bad, manual re-trigger is a pain, and delays current ETL (because of the single large queue of work to do). Make a daemon to that uses `etl.revision` to determine what's old and schedule for re-ETL. Do not deliver too much work at once. It can even look for missing data, and ETL that too.
You need to log in before you can comment on or make changes to this bug.