Backups & recovery

Backups & point-in-time recovery

ShipSquares provides continuous backups and point-in-time recovery (PITR) for managed PostgreSQL servers — physical base backups plus continuous write-ahead-log (WAL) archiving to S3-compatible storage, and an automated restore into a fresh, isolated container.

It works against any reachable Postgres with a replication-capable admin role and default settings — no server config change and no restart required, because the control plane captures over the network rather than editing postgresql.conf.

Capture

• Base backups on a schedule, streamed to your S3-compatible bucket, with keep-N and keep-N-days retention. Multi-tablespace clusters are supported.
• Continuous WAL archiving drains the write-ahead log on an interval through a dedicated replication slot that guarantees no gaps between runs. Your recovery point objective (RPO) is roughly the drain interval.

Restore

• Automated — restore into a fresh container, never touching the source. Recover to the end of the archived WAL, or to a specific point in time. The recovered instance comes up on its own port for you to inspect, pg_dump, or re-register.
• Manual runbook — generate the exact ordered steps (stop → wipe → extract → recovery config → start) to recover against an offline target yourself.

Good to know

• Backups require rclone and Docker on the control plane (the installer provisions rclone); automated restore recovers into a Postgres image matching the source’s major version.
• Client tool versions (pg_basebackup / pg_receivewal) must be at least the server’s version.
• A transaction committed in the not-yet-archived WAL segment is recovered once that segment drains — i.e. the RPO equals your archive interval.