Cluster version upgrade
When you change a cluster’s version, the cluster enters Updating and continues serving queries on the current version. On success it returns to Running on the new version; on failure it automatically reverts and returns to Running on the previous version, with the failure recorded in Activity History. See Zero-downtime upgrades. For In-VPC deployments, the cluster also exposes a deployment phase:Stable → Deploying → Switching → Draining → Cleanup → Stable (or RollingBack). See CloudPrem cluster operations.
Platform release states
A platform release (Platform Components) progresses through:
See Rollbacks, failure modes, and recovery and Status signals.