Add k8s 1.24 variants

[arnaldo2792]

Issue number:

Closes #2374

Description of changes:

sources: add symlink for metal-k8s-1.23 variant
sources: add symlink for vmware-k8s-1.23 variant
sources: add symlink for aws-k8s-1.23 variant
sources,variants: add metal-k8s-1.24 variant
sources,variants: add vmware-k8s-1.24 variant
sources,variants: add aws-k8s-1.24 variant
packages: add kubernetes 1.24

As part of this PR, the k8s 1.23 variants directories in sources/model/src/* were replaced with a symlink to the new k8s-1.24 variant for each platform.

Starting with k8s 1.24, all command-line parameters to manage CNI plugins with the kubelet were removed, as part of the efforts to clean up any remaining work related to the dockershim.

Testing done:

• Test aws variant: the nodes joined a 1.23/1.24 cluster , and pods were successfully deployed to them.

[ec2-user@admin]$ apiclient get os
{
  "os": {
    "arch": "x86_64",
    "build_id": "e08be563",
    "pretty_name": "Bottlerocket OS 1.10.0 (aws-k8s-1.24)",
    "variant_id": "aws-k8s-1.24",
    "version_id": "1.10.0"
  }
}
[ec2-user@admin]$ uname -a
Linux <ip>.us-west-2.compute.internal 5.15.54 #1 SMP Mon Sep 19 18:51:14 UTC 2022 x86_64 x86_64 x86_64 GNU/Linux

• Test vmware variant (Note: tested in a 1.23 cluster)
• Test metal variant (Note: tested in a 1.23 cluster)
• Build 1.23 variants
• Test ipv6 clusters

Terms of contribution:

By submitting this pull request, I agree that this contribution is dual-licensed under the terms of both the Apache License, version 2.0, and the MIT license.

[etungsten]

Remember to add and *-k8s.1.24* to the list of github actions workflows variant matrix as part of this PR.

[markusboehme]

There is a dependency between this PR and #2460. Whichever merges last will have to update the kubelet config template for k8s 1.24.

[stmcginnis]

Thanks for pointing that out. There's also work in #2377 that will need to be adjusted for this PR. I'll watch for this PR to go through and update accordingly, otherwise will make sure to help out here with any other changes.

[arnaldo2792]

(Forced push due to rebase onto develop)

[arnaldo2792]

Forced push includes:

• Adds aws-k8s-1.24-nvidia
  • This includes kubernetes.log-level from kubelet: Add setting to change log level #2460

[arnaldo2792]

(Rebase onto develop)

[kschumy]

The changes I suggested above are for the prod release we just cut. Conformance tests are running now. They're expected to pass in a few hours. I'll update y'all either way in the morning

[kschumy]

Conformance tests passed for EKS-D

[arnaldo2792]

( Forced push updates documentation and kubelet sources )

[arnaldo2792]

(Forced push fixes compatibility list in new k8s variants)

[jpmcb]

👏🏼

Testing done:

• Built the aws-k8s-1.24 variant with cargo make -e BUILDSYS_VARIANT=aws-k8s-1.24 and deployed an ami with cargo make -e BUILDSYS_VARIANT=aws-k8s-1.24 ami
• Deployed a k8s 1.23 cluster with the above ami via eksctl:

#!/usr/bin/env bash

eksctl create cluster \
    --name br-test \
    --version 1.23 \
    --region us-west-2 \
    --nodegroup-name br-ng \
    --node-type c5.xlarge \
    --node-ami ami-{ID}

• Nodes are up look good and via ssm I'm able to see the 1.24 version. I deployed the prometheues stack and see pods scheduled on the nodes.

[bcressey]

This doesn't seem correct - if this is happening we might not be building kubelet with the flags we expect.

[bcressey]

--network-plugin cni is no longer supported?

It'd be good to cover changes like this in the commit message, since the assumption otherwise is that nothing changed beyond the obvious stuff like versions.

[bcressey]

should be "1.24, 1.25, and 1.26"

[arnaldo2792]

( forced push adds more context to the message of the commit to add the new kubelet sources )

[arnaldo2792]

( forced push fixes docs and adds missing github actions for NVIDIA variant )

[arnaldo2792]

( forced push fixes comment about the missing debug symbols in generated binaries )