Updating a Node

Safe update order

1. Update the panel first (updating Catalyst). New agents expect a panel at least as new as they are.
2. Update one node and watch it for a full day before touching the rest.
3. Update remaining nodes in batches, never all at once.

Process per node

1. Announce maintenance — restarts interrupt running servers.
2. Trigger the agent update from Admin → Nodes (or your usual automation) and follow progress in the node view; the agent reports update stages back to the panel.
3. Verify: node healthy, a test server starts, console streams, backups list.
4. If the agent fails to come back, the node’s local logs and the previous agent binary are your recovery path — see node issues.

Rules

• Never update panel and all nodes simultaneously. A bad release then breaks everything instead of one canary node.
• Game servers themselves do not update with the agent — game versions come from templates (templates).
• Keep a changelog note per node (agent version + date) for the next incident.