Skip to main content
These answers cover the production database. For full reference, see Production databases and SQL database.
This error means the database compute endpoint is disabled. First, identify the affected database:
  1. Open the Database tool and select the database used by the failing app.
  2. If an Unpause database button appears, select it.
  3. Retry the connection.
If the endpoint keeps getting disabled, check Account usage and your billing settings for an unpaid invoice or usage limit. If the Unpause database control is missing, see “My database is frozen and I don’t see an Unpause button” before changing any Secrets.See Troubleshoot databases for the full diagnostic flow.
No. If the outage is already listed on status.replit.com, the team is working on it and a ticket won’t speed up the fix. Open the active incident and subscribe to updates to be notified as it progresses.If your database problem is not listed on the status page, contact Replit Support.
Replit supports point-in-time recovery for production databases. Available history depends on the configured recovery window; eligible plans can configure longer retention.A restore affects the whole database, not just the missing table, and can replace newer data. Review Data recovery before restoring. Increasing retention cannot recreate history that wasn’t kept.
First, find which database your app uses. A DATABASE_URL in Secrets may point to an intentionally configured external database.Compare the DATABASE_URL secret privately with the connection string in the Database tool. If your app should use the Replit-managed database but the secret points to an old connection, the secret is overriding the correct URL. Removing that stale override lets the app use the Replit-managed connection again. A different URL can also be intentional, so don’t remove it just because the values differ.As a last resort, if you’ve confirmed that the app should use its Replit-managed database and a stale DATABASE_URL secret is overriding that connection:
  1. Save the existing secret securely so you can restore it if needed.
  2. Delete only the conflicting DATABASE_URL entry from Secrets. Keep unrelated secrets and any intentional external database connection.
  3. Refresh, reopen the Database tool, and check for Unpause database again.
  4. Retry the app’s connection. If removing the secret breaks the connection, restore it.
If the control is still missing or the connection still fails, contact Replit Support with your project URL and checks completed. Don’t share connection strings.
Your app is opening more database connections at once than the pool allows. Common causes:
  • Connections that aren’t closed after each request
  • A traffic spike
  • Slow queries holding connections open
Most frameworks expect a single shared connection pool rather than a new connection per request.For the fastest fix, open a new Agent chat, describe the error, and ask Agent to check how your app manages database connections.
Replit keeps development and production databases separate so test changes don’t hit live data. When you publish, Replit identifies schema differences between your development and production databases and runs migrations automatically during build. To run a migration manually, select your production database in the Database tool and use the SQL console.Running SQL manually is an unsupported flow with limited support; review Apply missing objects to production before proceeding.See Create a production database when publishing and Troubleshoot databases.
Current billing doesn’t rule out a disabled endpoint. Follow the unpause steps above for the affected database.If unpause fails or the problem recurs, contact Replit Support with your project URL, error, and steps attempted.

​
Still need help?

If your database question isn’t answered here, contact Replit Support.