Operations — backup and restore¶
HA is not a backup. A deployment that was never restored from a backup has no backup. This page is Tier 0 — do it before anything else in this section.
What to back up¶
| Data | Where | How |
|---|---|---|
| Database (users, groups, plugin state, monitors, ITSM records, cluster tables) | MariaDB | mariadb-dump --single-transaction --routines --events lyndrix_db nightly; or mariabackup |
| Secrets and settings | Vault / OpenBao | vault operator raft snapshot save (Raft) or stop + tar of the file storage (file backend) |
| Unseal material | SECURITY_DIR/vault_keys.enc + LYNDRIX_MASTER_KEY |
copy the key file; keep the master key in your password manager — without both the Vault data is unreadable |
| Shared files | /data/storage |
file-level backup of the mount |
| Plugin artifacts | /app/plugins |
optional — they are re-installable from plugin_states (repo_url, desired_version) |
The cluster_* tables need no special handling; stale rows are pruned after a restore
(instances re-register on their next heartbeat).
Restore sequence¶
- Start Vault/OpenBao, restore the snapshot (or the file storage), unseal with the key file + master key (the core does this on boot when both are present).
- Restore the database dump into an empty
lyndrix_db. - Mount
/data/storagefrom backup. - Start one core instance in single mode first, log in, check Settings → Plugins (plugins re-install if the folder is missing) and the monitors/records.
- Switch to cluster mode and start the other instances.
Rehearse¶
Do the restore on a scratch host once per quarter and note the time it took: that is your real RTO. The dumps' age is your RPO.