A report that does not arrive is a failure too.
Backup software reports what happened. When a job stops running, it reports nothing, and nothing looks like a quiet good night. BackupSentinel knows when each report is due and flags the ones that never come.
Grace: 2 h 24 min for a daily job A check every 15 minutes
Silence gets a status of its own.
A job that was disabled, a server that was switched off, a notification setting that broke: none of them send an error, because the part that would send it is the part that stopped. So BackupSentinel does not wait to be told. Every job carries an expectation, and a job that misses it turns Missing, ranked with the failures.
Waiting and Paused are kept apart from Missing, so a job that has not reported yet, or a client you paused on purpose, does not raise a false alarm.
Every job knows when its next report is due.
Each job has an Expected every interval, from hourly to monthly. The next report is due one interval after the last one arrived.
- New jobs
- Count from the moment they are created. A job that was set up and never pointed at its address turns Missing too.
- Changed interval
- The clock restarts from the last report, so a job never turns late the moment you edit it.
- Suggested interval
- Once a job has five gaps between reports, its page suggests an interval from the typical gap.
- Paused clients
- No job turns Missing while its client is paused. Reports are still stored, and the clocks restart when it resumes.
A grace window that scales with the job.
A report that is a few minutes late is not a failed backup. Too little grace trains people to ignore alerts; too much finds real failures hours late. So the window grows with the interval:
- Every 8 hours or more often: a quarter of the interval, between 15 minutes and 2 hours.
- Longer intervals: a tenth of the interval, at least 2 hours.
Until the window closes, the job counts as late on the jobs list, and nothing is sent.
| Expected every | Grace | Missing after |
|---|---|---|
| Hourly | 15 min | 1 h 15 min |
| Every 2 hours | 30 min | 2 h 30 min |
| Every 3 hours | 45 min | 3 h 45 min |
| Every 4 hours | 1 h | 5 h |
| Every 6 hours | 1 h 30 min | 7 h 30 min |
| Every 8 hours | 2 h | 10 h |
| Every 12 hours | 2 h | 14 h |
| Daily | 2 h 24 min | 26 h 24 min |
| Every 2 days | 4 h 48 min | 2 d 4 h 48 min |
| Weekly | 16 h 48 min | 7 d 16 h 48 min |
| Monthly | 3 d | 33 d |
What happens when the window closes.
- Every 15 minutes, BackupSentinel checks every job against its due time and grace.
- A job past its window turns Missing and opens a critical alert: “missed its expected window”, with the interval it was expected at.
- If the job already had an open warning, that alert is raised to critical in place, and any acknowledgement or snooze is cleared so someone looks again.
- The alert goes to your channels, and the job goes to the top of Needs attention with the failures.
- When a good report arrives, the alert resolves itself and a recovery notice goes to the channels that received it.
What it does not do.
- Detection works on intervals, not calendars: there is no weekday-only schedule. Give a job the interval that covers its longest normal gap.
- The due time counts from when the report arrived, not from when the backup ran.
- The check runs every 15 minutes, so Missing can come up to 15 minutes after the window closes.
Find the backup that stopped reporting.
Point a backup's report at its client address, set how often it runs, and you will know the day it goes quiet.