Managed PostgreSQL
Create and connect PostgreSQL
Start with a small scale-to-zero database, then raise its compute ceiling when the application earns it.
Create the database
- Choose New resource, then PostgreSQL, or start with PostgreSQL during signup.
- Keep the 0.25 to 1 CU and five-minute sleep defaults, or open Advanced settings to change the active range and idle timeout.
- Create the database and copy DATABASE_URL when its stable connection identity is reserved; compute can finish preparing in the background.
Use the connection path
App bindings receive DATABASE_URL and POSTGRES_URL through protected references. Secret values remain out of ordinary portal state and logs.
The workspace owner can reveal and copy the pooled TLS verify-full connection for psql, application runtimes, and database clients. Never commit it to source control or paste it into support messages.
Run migrations deliberately
Your application or release process owns schema migrations, indexes, queries, and seed data. Test destructive changes before running them against production data.
A healthy database resource does not prove that the application schema is current. Verify a path that reads and writes the expected tables.
Plan and verify recovery
Encrypted logical backups can be restored into a separate database, leaving the source unchanged while you verify the recovered data.
Deleting a database first requires a retained encrypted backup. If one is not available, the same delete request creates recovery and continues automatically after the backup is safely cataloged.