Draft: Add CI jobs for testing rke2 node annotations

What does this MR do and why?

  • Introduces CI jobs, that are supposed to only run for RKE2 CI pipelines, that would test that all CP and MD nodes of the workload cluster have an expected test annotation - added initially because RKE2ConfigTemplate.spec.template.spec.agentConfig.nodeAnnotations does not result in any annotations natively and it's worked around within sylva-capi-cluster by RKE2ConfigTemplate.spec.template.spec.postRKE2Commands, which should change at some point and these jobs would allow for testing it, but please inform if you don't think we want them and I can drop the idea;

  • Does a change to the API commands used to fetch the Rancher kubeconfig in the test_no_sso, in 4a017a1e, by moving from jq to yq, simply because even though the default CI_IMAGE: registry.gitlab.com/sylva-projects/sylva-elements/container-images/ci-image:v1.0.15 has the binary now, in capm3-rke2-virt CIs inside Equinix, the jobs are based on a shell executor where the runner environment inside the Equinix server is defined in https://gitlab.com/sylva-projects/sylva-elements/ci-tooling/runner-aas/-/blob/main/user-data.sh.tpl?ref_type=heads, which already uses the mikefarah/yq, this way avoiding to introduce jq there.

Works on top of the change proposed in !1288 (merged).

Related reference(s)

Will probably need an evolution for the issue !1488 (closed) tries to solve, when connecting to the workload cluster inside libvirt-metal environments, unless a insecure flag for kubectl can be used.

Test coverage

Edited by Bogdan-Adrian Burciu

Merge request reports

Loading