Backups — Backup History & PITR Health
Backup recency, WAL archiver status, point-in-time recovery window.
What gets tracked
MonPG pulls backup metadata from your provider into one timeline. On RDS or Aurora it's automated and manual snapshots from describe-db-snapshots. On Azure Flex it's the automatic backups managed by the Azure Backup vault. On Cloud SQL it's daily automated plus on-demand via the Cloud SQL admin API. For self-hosted, MonPG can read pgBackRest, Barman, or WAL-G stanzas — point it at the backup catalog.
Health signals
Last backup age. If you're on a daily-backup config and the last successful backup is over 36 hours old, the page warns. Common cause is silent backup failure that nobody noticed because the previous backup is still there. WAL archiver lag. Pulled from pg_stat_archiver; a failing archiver means your PITR window is shrinking even if snapshots succeed. Backup size growth. Exponential growth in backup size is usually unbounded TOAST — check Schema Statistics. Restore tested? If you've never test-restored a backup, it doesn't count as a backup. We flag any backup not test-restored in the last 90 days.
Restore drill helper
"Suggest restore drill" opens a runbook tailored to your provider — exact CLI commands to spin up a clone instance from the latest snapshot, validate it, and tear it down. Quarterly cadence is reasonable. Most teams skip this and then learn the hard way that their backup process has been writing to a bucket they lost access to two months ago.