Backups and restore

Prolifto uses PostgreSQL. Backups are compressed pg_dump archives. Local backups go to backups/; production backups go to backups/prod/. Both are git-ignored.

Local development

pnpm db:backup
pnpm db:backup:verify
pnpm db:restore
pnpm db:restore backups/prolifto_dev-2026-10-25T12-00-00-000Z.dump --yes

• The most recent 14 days are kept by default.
• Override with DB_BACKUP_RETENTION_DAYS, DB_BACKUP_DIR, DB_BACKUP_SERVICE, DB_BACKUP_DATABASE, and DB_BACKUP_USER.
• db:restore takes a safety backup of the current database before restoring.
• Stop the app before restoring.
• Use --skip-backup only when the current database cannot be dumped.

Reset commands:

pnpm db:dev:reset        # takes a backup first
pnpm db:dev:reset:force  # no backup; use only for a broken database

Scheduled local backups on Windows

pnpm db:backup:install-task

The task waits for Docker and logs to backups/scheduled-backup.log. Change the time or name:

pnpm db:backup:install-task -Time 02:00 -TaskName "Prolifto Nightly Backup"

Remove it with:

pnpm db:backup:uninstall-task

On Linux/macOS, schedule pnpm db:backup with cron.

Production

pnpm db:prod:check
pnpm db:prod:backup
pnpm db:prod:migrate
pnpm db:prod:check

Production commands read DATABASE_URL from Vercel through vercel env run -e production. Backups require Docker.

• Run db:prod:check and db:prod:backup before a migration.
• db:prod:backup verifies the dump before reporting success.
• Run db:prod:check again after the migration.
• Set PROLIFTO_VERCEL_PROJECT if the Vercel project is not prolifto.

Restore drill

A backup is only useful if it restores. Periodically:

1. Take a fresh backup.
2. Restore it into a scratch database.
3. Run the app against the scratch database.
4. Verify a product, order, invoice, and customer.
5. Delete the scratch database.

Production retention

Use managed point-in-time recovery or pgBackRest/WAL-G with offsite storage. Keep an offsite copy separate from the application provider, and document who can restore it.

Related

• Database
• Deployment
• Observability