Datab Nodes update via snap channels. Each node opts into a channel (candidate for RC1) and receives updates within that channel automatically — with the snap system's atomic-rollback property: if the new revision fails to start, the snap reverts to the previous revision and the node keeps running.

How updates land

1. Snapd polls the channel for new revisions on the schedule defined per node.
2. If a new revision is available, snapd downloads it, validates the signature, and stages it.
3. Services restart against the new revision. Atomic — either all switch, or none do.
4. If any service fails its health check post-restart, snapd reverts to the previous revision.

What we do not do

• Silent downgrades from candidate to edge.
• Force-update during business hours without an opt-in window.
• Push updates with un-signed artifacts.

Pinning a node to a specific revision is supported. See Developer Docs for the operator commands.