Incident pattern

Supabase: project paused after a quiet week

Every database-backed page starts failing at once after a period of low traffic. Sign-in breaks, lists come back empty, and the dashboard shows the project as paused rather than reporting an error in your code.

  • Supabase
  • Lovable
  • Bolt
  • Auth & login
  • Forms & leads

Root cause, in plain English

Supabase pauses free-plan projects that see little activity across roughly a week, to reclaim server resources. The database stops accepting connections until someone restores it from the dashboard. Nothing in your application changed, which is what makes it so confusing: the last deploy was fine, the code is untouched, and the outage begins on a day with no traffic to trigger it.

How to fix it

  1. Check the project's state in the Supabase dashboard before debugging anything in your own code — a paused project explains every downstream symptom at once.

  2. Restore the project from the dashboard. Expect a short delay while the database comes back before requests succeed again.

  3. Decide whether this project should be on a paid plan. Anything a real customer touches should not be one quiet week away from a full outage.

  4. If it must stay on the free plan, keep it genuinely active with a small scheduled query rather than relying on organic traffic that may stop at any time.

  5. Add external monitoring that alerts on the outage. The pause is silent by design, so without a check you find out when a user tells you.

How Nightlamp detects this automatically

  • API canary
  • HTTP status
  • Browser journey

An api_canary calls a database-backed endpoint on a schedule, which both proves the database answers and counts as the activity that keeps a project from going idle in the first place. An http_status monitor catches the resulting 5xx on public pages, and a browser_journey confirms sign-in still completes rather than failing at the first query.

Catch this before your customers do

Nightlamp runs these checks continuously against your live app and sends a plain-English diagnosis — not a wall of logs — the moment this pattern shows up.

Frequently asked questions

How long does it take before a project pauses?
Supabase documents free-plan pausing against roughly a week of low activity. Treat it as approximate rather than a precise countdown — the practical takeaway is that a project nobody visits for several days is a candidate for pausing.
Is my data deleted when the project pauses?
Pausing is not deletion, and restoring brings the project back. That said, a paused project left indefinitely is not a backup strategy — keep your own exports of anything you would be unwilling to lose.
Does a scheduled ping definitely prevent it?
Regular real queries are what count as activity, so a scheduled check that actually reads from the database is far better than nothing. It is a workaround, not a guarantee — for anything customer-facing, a paid plan is the actual fix.

Newsletter

Get new incident patterns as we publish them

One email when new failure patterns, fixes, and monitoring recipes for no-code and AI-built apps land. No fluff, unsubscribe any time.

Double opt-in. One-click unsubscribe. No spam, ever.