Skip to content
Blog

Is Supabase down, or is it your project? How to tell in under a minute

Supabase's status page lists components per region but shows only what Supabase publishes. How to tell a platform incident from a problem in your project.

Your app starts throwing connection errors, queries time out, or auth stops issuing tokens. The first move is almost always the same: open a new tab, search "is Supabase down," and land on status.supabase.com. Ninety seconds later you are staring at a green board that says every component is operational, and the errors have not stopped.

The gap is not a bug in the status page. A status page has a specific job, and knowing what it does and does not cover gets you to an answer faster than refreshing the tab.

What does Supabase's status page track?

Supabase's status page tracks eleven components, each per region. The components are Compute, API Gateway, Database, Storage, Auth, Connection Pooler, Management API, Edge Functions, Realtime, Analytics, and the Dashboard. Regions include ap-northeast-1, eu-central-1 and us-east-1.

The per-region granularity is useful. A Database incident in one region does not have to turn the whole board red. If your project runs in us-east-1, the eu-central-1 row telling you nothing is exactly correct.

What the page does not show is how an incident gets onto it. Publishing a component as degraded or down is a decision, made by a person, after something has already been confirmed. Every status page shares that lag, not just Supabase's, and we covered the general pattern in why "is it down" sites disagree. If your project has been broken for ten minutes and the board is still green, that is not proof you are wrong. It might just be early.

Why does Supabase seem down when the status page is green?

Usually because the problem is in your project, not the platform. Before you file it as a platform incident, spend ten seconds each on three causes, roughly in order of how often they turn out to be the answer.

Cause On the status page? First check
Your own project (paused, exhausted pool, stale key, bad migration) No Dashboard logs, then supabase projects list
A regional incident outside your region Yes, in another region's row The row for your own region
A platform incident not yet published Not yet A probe from outside, with several regions

1. Your own project, not the platform. The most common cause of "Supabase seems down" is a problem scoped to one project. Four examples:

  • A paused free-tier project. Supabase's pricing page says Free projects are paused after 1 week of inactivity, and a paused project fails every request until it is restored, which reads exactly like an outage from the client.
  • A connection pool exhausted by an unbounded number of serverless function instances, each opening a new Postgres connection. Supabase's guide to connecting to Postgres explains the pooler options.
  • An expired or rotated API key.
  • A migration that changed a table your app still queries under the old shape.

None of these show up on any status page, because none of them are Supabase's problem.

Open your project's own logs first. The Supabase CLI does not ship a generic logs command; there is no supabase functions logs or supabase db logs. Logs live in the Dashboard under Logs, split into API, Postgres, and Edge Function logs. They are filterable by time range, status code, and path, and the Logs Explorer runs SQL over them for anything more specific than a quick scan.

From the CLI, supabase inspect db locks and supabase inspect db long-running-queries catch a stuck connection pool or a query holding a lock past what your app expects. The CLI reference documents both. supabase projects list confirms the project is not sitting paused in the first place. A specific error in any of these resolves the question immediately, usually faster than a probe from outside ever could.

2. A regional incident that does not affect your region. Suppose the status page shows Database degraded in ap-northeast-1 while you run in us-east-1. That is not your incident. Read the row for your own project's region, not the top-line banner.

3. An actual platform incident that has not been published yet. The status page structurally cannot help you with this case while it is happening. The incident only becomes visible once the lag closes.

How can you tell an unpublished Supabase incident from a project problem?

Run a check from outside both. You need a request from a location that is not your production servers, hitting a Supabase endpoint that is not scoped to your project.

A direct probe gives you what a status page and your own project logs cannot. An HTTP monitor against Supabase's public REST API (api.supabase.com/v1/projects) is one: a real, unauthenticated request from several regions, independent of any single project's configuration, with each region's reading shown separately.

A reading is one measurement from one place at one time. A probe can be wrong too, so treat a reading as a clue, not a verdict. If that endpoint answers while your project throws errors, that may point toward your project or your region, though an answering endpoint does not show Supabase is healthy. If it is failing from several regions, you have a timestamped reading (which region, when) to include when you open a support ticket.

The distinction changes what you do next:

  • "The platform is down" means wait, and maybe queue writes locally if your app can tolerate it.
  • "My project is broken" means the fix is in your own dashboard right now. Every minute spent refreshing status.supabase.com instead of checking your connection pool size is a minute your users spend on a broken app that Supabase cannot fix for you.

How do you monitor a Supabase-backed app itself?

Monitor the thing your users hit, not the platform underneath it. A platform status page, however good, answers "is Supabase up," never "is my app up." Your app can go down while Supabase is perfectly healthy, through a bad deploy, a migration that broke a query, or an environment variable that did not carry over. A platform-level check shows green the entire time.

On RealUptime Monitoring, three checks cover the app:

  • An HTTP monitor against your app's real URL (not Supabase's), from every region your plan includes, on the interval you choose.
  • A TCP monitor, if you connect directly to the Postgres instance over the pooler. It tells you the moment the port stops accepting connections, independent of whether a query would succeed.
  • A heartbeat monitor for anything that runs on a schedule against your Supabase project, such as a cron job doing a nightly export or a queue worker consuming a Postgres LISTEN/NOTIFY channel. It alerts on silence rather than on a failed request, and our heartbeat monitoring guide covers it in detail.

None of this requires access to Supabase's infrastructure or an enterprise support plan. Each check points at a URL you already have and takes a few minutes to set up.

Can you get notified without refreshing a status page?

Yes, with a check of your own. Checking by hand, in the moment, is still a step slower than being told, for a service your product genuinely cannot run without.

The HTTP monitor on Supabase's public REST API described above alerts through your own channels when that endpoint stops answering from several regions. It tells you when Supabase's API changes state; your app's checks tell you when your app stops working, regardless of whose fault it is. Green on status.supabase.com means only that Supabase has not yet confirmed a problem, so keep both in place.