Guide · Backups

Monitor backups with a dead man's switch

A backup job that crashes loudly is easy to catch. One that quietly stops running, because a cron entry got deleted, a credential expired, or a host was rebuilt without it, is invisible until the day you need to restore. One check-in call closes that gap.

1. Create a check

Create a check in Mortemain with your backup's schedule and grace window, then copy its unique ping URL:

your ping URL

https://ping.mortemain.com/your-check-uuid

2. Check in when the backup finishes

Add one request to the end of the job, reached only if everything before it succeeded. A plain GET or POST to the base URL is enough:

backup.sh

#!/bin/sh
CHECK_URL="https://ping.mortemain.com/your-check-uuid"

# ... your backup command(s) here ...
/usr/local/bin/run-backup.sh && \
  curl -fsS -o /dev/null "$CHECK_URL"

3. Alert immediately on failure

Don't just wait for the check-in to go missing, tell Mortemain the moment the job fails so you're paged straight away instead of at the end of the grace window:

backup.sh

if /usr/local/bin/run-backup.sh; then
  curl -fsS -o /dev/null "$CHECK_URL"
else
  curl -fsS -o /dev/null "$CHECK_URL/fail"
fi

4. Optional: track how long it takes

Ping /start right before the backup begins and the base URL when it finishes. Mortemain then measures the run's duration, so a backup that starts on time but silently hangs still gets caught:

backup.sh

curl -fsS -o /dev/null "$CHECK_URL/start"
/usr/local/bin/run-backup.sh && curl -fsS -o /dev/null "$CHECK_URL" \
  || curl -fsS -o /dev/null "$CHECK_URL/fail"

5. Pick your backup tool

The pattern above is the same everywhere: check in on success, ping /fail on error. Here's the exact hook for the tools we get asked about most:

PostgreSQL (pg_dump)

Wrap pg_dump and check in on a clean exit code.

rsync

Check in only when the sync itself exits clean.

BorgBackup

Ping after backup and prune, not just the first step.

Restic

Check in on backup, and on forget/prune too.

Duplicati

Use its run-script hooks to check in without touching the job.

Synology Hyper Backup

A GUI-only tool, wired up with its notification script hook.

TrueNAS

Post-script hooks on Cloud Sync and replication tasks.

Veeam

A post-job script that proves the backup actually finished.

Test it

Run the backup once: the check flips to up within a second or two. Then disable the job (or comment out the cron line) and wait past the grace window, you should get a down alert with nothing else needed. That's the whole point: the alert fires because nothing happened, not because something crashed.