Upgrade
The Upgrade tab upgrades the cluster's Kubernetes (K3s / RKE2) version, updating nodes one at a time safely.
Related: Kubernetes
Three steps
| Step | Content |
|---|---|
| 1. Version check | Review the current version, the latest available version, and per-node status |
| 2. Plan (dry run) | Preview the steps that will run on each node — no changes are made to the cluster |
| 3. Apply | Actually run the upgrade with real-time progress |
1. Version check
- Shows the current version, the latest available version, and how many minor versions behind you are.
- Suggests a recommended upgrade (e.g. a patch within the same minor — lowest risk, no API changes).
- Pick a target version and continue to Review plan.
2. Plan (dry run)
For each node to be upgraded, shows the steps that will run (cordon → drain → ssh-upgrade → wait-ready → uncordon). No changes are made at this stage.
Options
| Option | Description |
|---|---|
| Drain timeout (s) | How long to wait for a node to drain |
| Ready timeout (s) | How long to wait for a node to become Ready after upgrade |
| Skip drain | Proceed without draining |
| Halt on drain failure | On: stop the whole run if a drain fails / Off (default): continue best-effort |
Before running, you must acknowledge that nodes are drained and rebooted one at a time.
3. Apply
- Real-time overall progress (completed node count) and per-node step status.
- The event log shows the timeline:
start/cordoned/drained/upgrading/failed/uncordoned, etc. - On failure: the failed node is left cordoned. After investigating, Retry or uncordon it manually (it may be auto-uncordoned depending on the settings). You can change the drain options on retry.
Safe by design: nodes are processed one at a time, and the run stops on failure rather than continuing.
On this page
Three steps
1. Version check
2. Plan (dry run)
Options
3. Apply