Video · 40 sec
A failed sync, reported in the group
The design
- A scheduled job in the bot queries the last sync run and the drift view every hour, offset from the sync itself.
- Three alert types: no successful run in N hours, a run with an error, and drift above tolerance on any account.
- Each alert carries the fix from the runbook and is suppressed until the condition changes, so the group is never spammed.
What I expect to learn
The right stale threshold. Three hours is the current guess in the settings row.