-
Sub-task
-
Resolution: Unresolved
-
Undefined
-
None
-
None
-
None
-
Quality / Stability / Reliability
-
False
-
-
False
-
-
-
Installer Sprint 2025-68, Installer Sprint 2025-69, Installer Sprint 2025-70
Test cases
- The hypershift and hypershift-local-hosting components are enabled by default and cluster-api and cluster-api-provider-aws components are disabled by default in MCE. Ensure that I can enable cluster-api and cluster-api-provider-aws components only if the hypershift components are disabled.
- When cluster-api and cluster-api-provider-aws components are enabled, ensure that the components work.
- Disabling the hypershift components leave the CAPI and CAPA related CRDs behind.
- Enabling cluster-api and cluster-api-provider-aws components will upgrade the remaining CRDs from the disabled hypershift components so CAPI and CAPA will work without having to manually clean up the CRDs after disabling hypershift components but before enabling the CAPI/CAPA components.
- Verification: do some CAPI/CAPA related operations like provisioning a cluster with it.
- Continuing from TC2, when a user wants to disable CAPI/CAPA and re-enable the hypershift components,
- Disabling the CAPI/CAPA components leave the CAPI and CAPA related CRDs behind.
- Re-enabling the hypershift components tries to install (downgrade) its version of CRDs. Downgrading CRDs is not allowed in kubernetes. Therefore, the hypershift operator will fail its hosted cluster operations.
- Verifications:
- Check that the hypershift operator fails its hosted cluster operations such as creating a new HCP cluster.
- Removing the CAPI/CAPA related CRDs before re-enabling the hypershift component resolves the problem.
- We will have to document this case and which CRDs to clean up in this flow.
- Similar to TC3, I have MCE with CAPI/CAPA components enabled. Now I uninstall MCE and re-install MCE.
- Uninstallation with CAPI/CAPA enabled will leave their CRDs in the cluster.
- Re-installing MCE will enable hypershift components by default and CAPI/CAPA disabled. In this flow, the hypershift operator will fail because of the reason in TC3.
- Verifications:
- Check that the hypershift operator fails its hosted cluster operations such as creating a new HCP cluster.
- Removing the CAPI/CAPA related CRDs before re-installing MCE resolves the problem.
- We will have to document this case and which CRDs to clean up in this flow.