Changes:
- Rancher version convention was changed from `v2.10.0` to `2.10.0` to better map to the actual version in the upstream helm chart repo which is without the `v` prefix.
- Rancher's `postDelete` hook has been disabled - done to ensure that we will not hit a corner case where:
1. The Rancher helm chart upgrade fails, because of a core component not yet being ready
2. The `helm-controller` schedules a `helm uninstall` which deletes the Rancher Helm release and triggers the `postDelete` hook.
3. The problematic core component is up and running, so `helm-controller` schedules a `helm install` with the new version.
4. Due to insufficient resources, or network connection (or other unforeseen problems), the `postDelete` hook is still running and it wrongly removes the new Rancher installation resulting in a missing rancher from the cluster after an upgrade.
The `postDelete` hook ensures that no accidental delete of the Rancher application will happen during an upgrade over a machine with fewer resources.
Reviewed-on: suse-edge/Factory#60
Reviewed-by: Denislav Prodanov <dprodanov@noreply.src.opensuse.org>
Reviewed-by: Nicolas Belouin <nbelouin@noreply.src.opensuse.org>
Reviewed-by: Atanas Dinov <atanasdinov@noreply.src.opensuse.org>
Co-authored-by: Ivo Petrov <ivo.petrov@suse.com>
Co-committed-by: Ivo Petrov <ivo.petrov@suse.com>
- Bumps both RKE2 and K3s versions to the `1.31.3` version that is expected by Rancher `v2.10.1`.
- Bumps the K8s core component versions to the `1.31.3` expected versions.
RKE2 core component versions have been checked against the `Chart Versions` table of the said [release](https://github.com/rancher/rke2/releases/tag/v1.31.3%2Brke2r1).
K3s core component versions have been checked agains the [manifests](https://github.com/k3s-io/k3s/tree/v1.31.3%2Bk3s1/manifests) directory of said release.
Reviewed-on: suse-edge/Factory#58
Reviewed-by: Denislav Prodanov <dprodanov@noreply.src.opensuse.org>
Co-authored-by: Ivo Petrov <ivo.petrov@suse.com>
Co-committed-by: Ivo Petrov <ivo.petrov@suse.com>
Introduces the K8s distribution core component list that the upgrade-controller will follow in order to make sure that a specific Kubernetes upgrade has completed successfully.
Relates to the [#116](https://github.com/suse-edge/upgrade-controller/pull/116) upgrade-controller PR.
Reviewed-on: suse-edge/Factory#52
Reviewed-by: Atanas Dinov <atanasdinov@noreply.src.opensuse.org>
Reviewed-by: Nicolas Belouin <nbelouin@noreply.src.opensuse.org>
Co-authored-by: Ivo Petrov <ivo.petrov@suse.com>
Co-committed-by: Ivo Petrov <ivo.petrov@suse.com>
The previous import was based on a pre-merge copy of the following PR
- an issue was discovered during SV validation which required an
additional change to ensure CRDs are created before creating the
ClusterctlConfig CR
https://github.com/suse-edge/charts/pull/166
After further testing I discovered that the cluster-api-controller is
not correctly pinned or using the downstream image, and a similar
problem exists for CAPM3 (but only after upgrade from an older chart)
due to a mistake in the templating.
