New
-
After a successful
create, the tool prints the command that uses the kubeconfig it just wrote:Your cluster is ready! To access it, run: export KUBECONFIG="/path/to/kubeconfig" kubectl get nodesThe path is the expanded
kubeconfig_path, so the command works from any directory, and it is quoted so a path with spaces survives a copy and paste. A cluster created with an older version gets the same message from its nextcreate. - by @qaiser42 in #789 -
New etcd Snapshots and Restore page: how the hourly snapshots are configured, where to find them, and how to restore from a local file or an S3 bucket on single and multi-master clusters.
Improvements
- Every copy-paste configuration on the docs site carries
config_format_versionnow: the README quick start, the complete tutorial, and both samples on the recommendations page. Copying any of them no longer stops at the format check. - by @qaiser42 in #789 - The tutorial notes that
createprints the kubeconfig export line at the end.
Fixes
- The suite covers the three Hubble toggles:
hubble_enabled,hubble_relay_enabledandhubble_ui_enabledhonour an explicitfalse, and Relay and UI follow Hubble unless they are set. The code fix shipped in v2.12.0, and this release adds the spec that keeps it in place. - by @recepdmr in #807
Upgrading
No configuration change is required.
- Existing clusters are unaffected by the new output. The next
createprints the access instructions at the end and re-runs the installation as before. - Cilium clusters: a Hubble toggle set to
falsetakes effect on the nextcreate, which re-renders the values and runshelm upgrade. - Bundled addon pins are unchanged since v2.17.0: the cloud controller manager, the CSI driver, the system upgrade controller and the cluster autoscaler keep their versions.
If hetzner-k3s is useful to you or your company, please consider sponsoring its development.
Sponsorship helps ensure continued maintenance and new features. Thank you to our current sponsors! 🙏