⚠️ Use k8s v1.27.0 in quickstart docs and CAPD

[furkatgofurov7]

Signed-off by: Furkat Gofurov (furkat.gofurov@suse.com)

What this PR does / why we need it:
Bumps k8s version to v1.27.0 in Quickstart docs and CAPD.

NOTE:

• etcd version is bumped to 3.5.7-0 following the kubeadm
• COREDNS_VERSION is bumped to v1.10.1 following the kubeadm
• new field called: WorkloadKubernetesVersion was introduced to ClusterctlUpgradeSpecInput

Which issue(s) this PR fixes (optional, in fixes #<issue number>(, fixes #<issue_number>, ...) format, will close the issue(s) when PR gets merged):

Part of: #8459

cc @kubernetes-sigs/cluster-api-release-team

[furkatgofurov7]

/hold

Manually verifying quickstart is pending

[furkatgofurov7]

/hold

Manually verifying quickstart is pending

Done and all good.

/hold cancel

[furkatgofurov7]

/assign @sbueringer

[furkatgofurov7]

GH rate limit issue

/test pull-cluster-api-e2e-informing-ipv6-main

[sbueringer]

Just a few nits

[sbueringer]

/test pull-cluster-api-e2e-full-main

^^ this test has to be green before merge

[sbueringer]

/retest

[killianmuldoon]

/retest

possibly a flake

[sbueringer]

/retest

[sbueringer]

@furkatgofurov7 @killianmuldoon Let's wait with test-infra PR merges until this PR is green

[killianmuldoon]

Definitely consistent failure in the 0.3 upgrade test 🤔. I'll have a look certainly by tomorrow some time (could have been a good candidate for the hacking session 😄)

[furkatgofurov7]

/test pull-cluster-api-e2e-full-main

[furkatgofurov7]

/retest

[sbueringer]

Ups, is this test actually trying to create a 1.27 workload cluster with CAPI v0.3?

INFO: Creating the workload cluster with name "clusterctl-upgrade-eq86nx" using the "(default)" template (Kubernetes v1.27.0, 1 control-plane machines, 1 worker machines)

If I see this correctly we should introduce a new parameter to the test spec for the test cluster kubernetes version

[sbueringer]

Context:

• This test creates a separate management cluster with Kubernetes version: InitWithKubernetesVersion
• Then the old providers are deployed on this cluster
• Then we create a test workload cluster via the old providers

Proposal:

• I would suggest to introduce a new field to ClusterctlUpgradeSpecInput called: WorkloadKubernetesVersion
• This field should be set to the maximumg workload cluster version supported by the old providers

So we end up with the following:

Test	InitWithKubernetesVersion	WorkloadKubernetesVersion
v0.3	v1.21.14	v1.22.17
v0.4	v1.23.17	v1.23.17
v1.0	v1.23.17	v1.23.17
v1.3	v1.26.3	v1.26.3
v1.4	v1.27.0	v1.27.0

InitWithKubernetesVersion is the version we deploy the old provider on
WorkloadKubernetesVersion is the version of the workload cluster the old provider creates

Both those version should be the highest the old provider supports (this makes it very likely that the main version also supports those versions, and as of today that always works)

Note: patch version should be the highest we have corresponding kindest/node image for on dockerhub. I looked them up and "bumped" them a bit. We don't have to continuously keep them up-to-date, but maybe let's do it now as we have to introduce WorkloadKubernetesVersion anyway

@furkatgofurov7 @killianmuldoon @fabriziopandini @ykakarap Opinions?

[furkatgofurov7]

Ups, is this test actually trying to create a 1.27 workload cluster with CAPI v0.3?

yeah, but that is the case with 0.4, as well no? And I saw that went fine 🤔

[sbueringer]

Ups, is this test actually trying to create a 1.27 workload cluster with CAPI v0.3?

yeah, but that is the case with 0.4, as well no? And I saw that went fine thinking

I could imagine that v0.4 is for some reason compatible with v1.27 (e.g. we use a newer client-go, controller-runtime version).

But it should be relatively easy to see by running the failing test locally and just looking why the Machines are not coming up (I just don't have the time for it right now).

I think we should definitely find out why the test is actually failing.

I would do the suggested change in any case btw, even if we don't need it for this specific PR. It might just be the solution we need if we find out that v0.3 is simply not compatible with Kubernetes v1.27

[furkatgofurov7]

Running it once again to be sure nothing new was introduced while rebasing:

/test pull-cluster-api-e2e-full-main

[killianmuldoon]

Looking good! Thanks for getting this one over the line - was a bit more complex than usual 🙂

/lgtm

[k8s-ci-robot]

LGTM label has been added.

Git tree hash: ea2f0bacea383ae5e42a5af387f2a412defa62f7

[sbueringer]

Thank you very much!!

/approve

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: sbueringer

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [sbueringer]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[furkatgofurov7]

@killianmuldoon @sbueringer thx folks, do we need to back-port it to release-1.4 branch?

[sbueringer]

@killianmuldoon @sbueringer thx folks, do we need to back-port it to release-1.4 branch?

Yes, please!

I think we can & should definitely prepare the PR already. Then we have to decide if we want to "officially" support 1.27 already in 1.4.2 or merge it intentionally afterwards to get more time in CI

[furkatgofurov7]

Yes, please!

I think we can & should definitely prepare the PR already. Then we have to decide if we want to "officially" support 1.27 already in 1.4.2 or merge it intentionally afterwards to get more time in CI

ack, let me try the cherry-picker and see if it can do it otherwise will manually cherry pick it:

/cherry-pick release-1.4

[k8s-infra-cherrypick-robot]

@furkatgofurov7: #8518 failed to apply on top of branch "release-1.4":

Applying: Bump k8s version to v1.27.0 in quickstart docs and Tiltfile
Applying: Use k8s v1.27.0 in CAPD and tests
Using index info to reconstruct a base tree...
M	test/e2e/clusterctl_upgrade_test.go
M	test/e2e/config/docker.yaml
Falling back to patching base and 3-way merge...
Auto-merging test/e2e/config/docker.yaml
Auto-merging test/e2e/clusterctl_upgrade_test.go
Applying: Introduce WorkloadKubernetesVersion field to ClusterctlUpgradeSpecInput
.git/rebase-apply/patch:44: trailing whitespace.
- Providers running CAPI release-0.3 clusterctl upgrade tests should set `WorkloadKubernetesVersion` field to the maximum workload cluster kubernetes version supported by the old providers in `ClusterctlUpgradeSpecInput`. For more information, please see: https://github.com/kubernetes-sigs/cluster-api/pull/8518#issuecomment-1508064859 
warning: 1 line adds whitespace errors.
Using index info to reconstruct a base tree...
A	docs/book/src/developer/providers/migrations/v1.3-to-v1.4.md
A	docs/book/src/developer/providers/migrations/v1.4-to-v1.5.md
M	docs/release/release-tasks.md
M	test/e2e/clusterctl_upgrade_test.go
Falling back to patching base and 3-way merge...
Auto-merging test/e2e/clusterctl_upgrade_test.go
CONFLICT (content): Merge conflict in test/e2e/clusterctl_upgrade_test.go
Auto-merging docs/release/release-tasks.md
Auto-merging docs/book/src/developer/providers/v1.3-to-v1.4.md
CONFLICT (modify/delete): docs/book/src/developer/providers/migrations/v1.4-to-v1.5.md deleted in HEAD and modified in Introduce WorkloadKubernetesVersion field to ClusterctlUpgradeSpecInput. Version Introduce WorkloadKubernetesVersion field to ClusterctlUpgradeSpecInput of docs/book/src/developer/providers/migrations/v1.4-to-v1.5.md left in tree.
error: Failed to merge in the changes.
hint: Use 'git am --show-current-patch=diff' to see the failed patch
Patch failed at 0003 Introduce WorkloadKubernetesVersion field to ClusterctlUpgradeSpecInput
When you have resolved this problem, run "git am --continue".
If you prefer to skip this patch, run "git am --skip" instead.
To restore the original branch and stop patching, run "git am --abort".

In response to this:

Yes, please!

I think we can & should definitely prepare the PR already. Then we have to decide if we want to "officially" support 1.27 already in 1.4.2 or merge it intentionally afterwards to get more time in CI

ack, let me try the cherry-picker and see if it can do it otherwise will manually cherry pick it:

/cherry-pick release-1.4

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[sbueringer]

No chance with the migration doc :D

[furkatgofurov7]

No chance with the migration doc :D

Here: #8538

[johannesfrey]

/area provider/infrastructure-docker e2e-testing documentation