Self-hosting

Backups, restore and moving machines

Take an online backup of the single SQLite file, copy it off the machine, restore it safely, and move Glideworth to a new machine.

All your data is one SQLite file on the glideworth-data volume. A backup is one self-contained file, is integrity-checked, and can be taken while the server is running.

Back up

# A named file on the volume
docker compose exec glideworth node dist/cli/admin.js backup --out /data/backups/manual.db

# Or let it name the file: /data/backups/<database name>-<timestamp>.db
docker compose exec glideworth node dist/cli/admin.js backup

# Copy it off the volume to the host (and then off the machine)
docker compose cp glideworth:/data/backups/manual.db ./glideworth-backup.db

Without Docker, from a checkout: pnpm backup, or pnpm backup -- --out ~/glideworth.db.

Backups on the same volume protect against mistakes, not against losing the disk: copy them somewhere else, regularly. For a daily backup, cron on the host can run the first command with a dated name and docker compose cp, followed by an off-site copy (rsync, restic or rclone). In cron, add -T to docker compose exec, which has no terminal there.

You can also export one household as portable JSON from the app, or with GET /households/{id}/export. That is not a full backup: it has no users, keys or history.

Restore

Stop the server first, so nothing writes while the file is replaced. Then:

docker compose stop glideworth

# Backup file is on the host, in the current directory:
docker compose run --rm --no-deps \
  -v "$PWD/glideworth-backup.db:/restore/backup.db:ro" \
  glideworth node dist/cli/admin.js restore --from /restore/backup.db --force

docker compose up -d

Without Docker: stop the server, then pnpm restore -- --from ~/glideworth.db --force.

--force is required when a database already exists. The current database is kept next to it as <database>.pre-restore-<timestamp> on the volume, so a wrong restore can be undone. Delete those copies when you no longer need them. A backup from an older version is migrated to the current schema.

Moving to a new machine

  1. On the old machine, take a backup and copy it off. Also copy your .env file: it holds the setup token and settings, and is not inside the backup.

  2. On the new machine, install Docker, clone the repository (the same version or newer), put the .env in place, and build without starting:

    docker compose build
  3. Restore the backup into the new, empty volume before the first start. No --force is needed because no database exists yet.

    docker compose run --rm --no-deps \
      -v "$PWD/glideworth-backup.db:/restore/backup.db:ro" \
      glideworth node dist/cli/admin.js restore --from /restore/backup.db
    docker compose up -d
  4. Sign in with your existing passphrase. Users, API keys and share links travel with the database; existing browser sessions may need to sign in again. Update DNS and the proxy, and set GLIDEWORTH_PUBLIC_URL if the address changed (assistants connected over OAuth will need to reconnect).

  5. Keep the old machine’s data until you are sure the new one works, then delete the old volume.

Last checked against the code on .