Maintenance

Always back up the configuration (especially its secret key), database, uploaded images, local themes, and extensions. Treat exports as sensitive: they contain personal data, password hashes, and tokens.

Backup and restore#

For a consistent application JSON export, stop web and worker processes first, then run:

quillsail --config /var/lib/copydesk/copydesk.toml export /secure/backup/copydesk.json

Copy the instance configuration, uploads directory, themes, and extensions to protected storage. For native database backups use the database’s consistent backup facility. Do not simply copy a running SQLite database while ignoring its WAL files.

To restore, prepare an isolated empty database and configuration using the same QuillSail version and secret key. Do not run init on the restore target because it creates an administrator. Create the configured data directory (and any parent directory for a SQLite database file). Point the configuration at the empty database, then run:

quillsail --config /path/to/restored/copydesk.toml import /secure/backup/copydesk.json

Copy uploads, themes, and extensions to the configured locations. Verify sign-in, published content, subscriptions, and the queued-job list before starting the worker. Restoring a backup can restore pending jobs: inspect these before enabling SMTP to avoid sending old mail. Keep the original backup until recovery has been tested.

Upgrade#

For a registered Linux Python/systemd or Compose installation, run sudo quillsail update. It prepares the authenticated release, pauses traffic and jobs, backs up the complete instance and native database, migrates, activates and checks web/worker health, and recovers failures. See server updates for one-time setup, older-server bootstrap, --check, --resume and rollback. Desktop installations use the launcher’s update button or quillsail update --app PATH.

For deployments managed by another orchestrator, follow its maintenance process:

Stop web and worker processes, take a backup, install the new package, run quillsail upgrade with the instance configuration, then restart services. Test on a restored staging instance first. Migrations run through Alembic; downgrades are not a supported operational recovery mechanism. Roll back by restoring the previous package and its matching database backup.

Check /healthz for database reachability, the dashboard System page for worker activity, and service logs for errors. Health checks do not prove SMTP or worker availability. quillsail check, quillsail jobs list, quillsail user --help, and quillsail subscribers --help provide maintenance commands. Keep dependency updates and restore rehearsals on the organization’s maintenance schedule.