Upgrades and recovery
What to preserve
Game snapshots and installation recovery protect different data. Preserve:
- The panel database and its persistent state directory, including the Spark signing authority and encrypted integration credentials
- Each Spark’s identity, runtime metadata, game data, and backup archives
- Any separately configured S3 backups and their retention settings
Before upgrading, stop the panel and take a consistent database and state backup. For SQLite, copying only a live database file can omit WAL contents. For PostgreSQL, back up the database as well as the panel state directory.
Discord delivery identities are stored in the database, while the token files
and encryption key live in integration-secrets under the panel state directory.
Restore them together; see Discord backup and recovery.
Docker Compose upgrades
Select the desired image tag in your deployment configuration, then run:
docker compose pulldocker compose up -dCheck login, health, and Spark connections afterward. Do not use
docker compose down -v during upgrades: it deletes persistent volumes.
Binary upgrades
Stop the relevant service, replace its binary with the verified release, and restart it. Preserve the previous binary until verification succeeds.
Upgrade Spark when required by protocol or feature compatibility. Wait for active operations to finish before restarting an agent.
Recovery and rollback
Database schema checks run at startup. A rollback may require restoring the matching pre-upgrade database and state as well as the previous image or binary.
History retention when upgrading
Audit and completed operation history are retained for 90 days by default.
Records older than the configured retention period are eligible for automatic
deletion. To keep them indefinitely, set both
CAMPFIRE_AUDIT_RETENTION_DAYS=0 and CAMPFIRE_OPERATION_RETENTION_DAYS=0 in the
panel environment. Before upgrading, check that your retention settings match
the history you want to preserve. See the
settings reference.
Administrator password recovery
Save a new password of at least 15 characters in a private, mode-0600 regular file. On a binary installation, run:
sudo campfire reset-password \ --database /var/lib/campfire/panel.db \ --password-file /private/new-passwordFor Docker, place the file in the container’s writable data directory, readable only by its UID 10001, then run:
docker compose exec campfire campfire reset-password \ --password-file /var/lib/campfire/new-passwordUse your configured database path or PostgreSQL URL for binary installations. Add --username NAME to select an administrator; otherwise the first administrator is recovered. Recovery revokes that account’s sessions. Remove temporary password files afterward. If local login is disabled, re-enable it in your deployment configuration before signing in with the recovered password.