Acceleratio v0.3.1 is outRelease notes →
Skip to Content
AcceleratioDocs

PostgreSQL

Databases with continuous backups and point-in-time restore.

Create a database with a version (16 or 17), a disk size and a storage profile. It runs as one CloudNativePG instance.

The Create PostgreSQL dialog: name postgres, PostgreSQL 17, 10 GiB

A database is a single instance without automatic failover. Its disk has one copy on one server unless its storage profile keeps more.

Connect

The PostgreSQL overview: connections, database size, cache hits, transactions and the connection string

The overview shows the in-project connection address and, on request, the password. Services of the same project connect over the private network; a database needs no public domain.

The overview also shows the server's state and load, and the tables of the app database with their indexes and a read-only, paged view of their rows.

Access from outside the cluster

The database port is closed to the outside by default. Settings can open it as a node port (30000–32767) on the server that runs the database, limited to listed addresses or networks, or explicitly to everyone.

  • Connections from outside must use TLS (sslmode=require). The certificate is the cluster's own, so clients cannot verify it yet.
  • The address follows the server, so it can change when the database moves.

Manage

  • Stop (hibernate), start and restart the database.
  • Grow the disk from Settings, without a restart, when the storage profile allows volume expansion. A disk cannot shrink.
  • Regenerate the password after a confirmation. The database switches to it within seconds without touching data; update services that use the old password by hand.
  • Delete the database, which deletes its disk, its data and its backups.

The canvas card shows the used and free space of the disk, with a warning from 85%.

Backups

Every database is backed up continuously:

  • its WAL is archived as it is written, and sent when a segment fills or every 5 minutes;
  • a base backup runs daily at 03:00 UTC, and on demand from the Backups tab;
  • backups and WAL are kept 7 days, in a bucket only this database's keys can reach.
Continuous backups and point-in-time restore

The Backups tab: backups on, point-in-time recovery active with 7 days retention, a completed base backup, export and import

Restore creates a new database in the same project from any moment since the first backup, with a new password. The source keeps running.

Backups are stored on the same servers as the database. They protect against mistakes such as a dropped table, not against losing the servers' disks. A restore after losing the server can miss the last few minutes.

Backups are on by default. Turning them off, after a confirmation, stops archiving and deletes every backup. Turning them on again starts from a new base backup.

Export and import

  • Export downloads the app database as it is now, as a pg_dump custom-format file. Restore it with pg_restore of the same PostgreSQL version or newer.
  • Import loads a dump as the app user, all or nothing in one transaction. A custom-format archive replaces the objects it contains. A SQL file runs as written, so make it with pg_dump --no-owner --no-acl.

Not supported yet

Backups to an external S3 store, a custom backup schedule or retention, restoring over the existing database, replicas, failover and connection pooling.

Was this page helpful?

Last updated on

On this page