RepoJournal
Kubernetes

@kubernetes

Container orchestration — what platform teams ship on

Pick a date

  1. Thu 21 may
  2. Sat 23 may
  3. Sun 24 may
  4. Mon 25 may
  5. Tue 26 may
  6. Wed 27 may
  7. Thu 28 may
  8. Fri 29 may
  9. Sat 30 may
  10. Sun 31 may
  11. Mon 1 jun
  12. Mon 8 jun
  13. Tue 9 jun
  14. Wed 10 jun
  15. Thu 11 jun
  16. Fri 12 jun
  17. Sat 13 jun
  18. Sun 14 jun
  19. Mon 15 jun
  20. Tue 16 jun
  21. Wed 17 jun
  22. Thu 18 jun
  23. Fri 19 jun
  24. Sat 27 jun
  25. Sun 28 jun
  26. Mon 29 jun
  27. Tue 30 jun
  28. Wed 1 jul
  29. Thu 2 jul
  30. Today 3 jul

The Wire · Showcase

KARPENTER LANDS IN KOPS, KUBE-PROXY CUTS CADVISOR DEPENDENCY

By RepoJournal · Filed · About Kubernetes

Kops now ships native Karpenter EC2NodeClass and NodePool support while kube-proxy eliminates its cadvisor dependency by reading CPU topology directly from cpuset.

Kops merged full managed Karpenter integration [1], giving operators declarative EC2NodeClass and NodePool configuration without leaving the kops cluster spec. This lands alongside command-line slice expansion [2], which now allows paths like `cluster.spec.addons[0].manifest` to create missing array elements on the fly, unblocking Gateway API CRD installation via the addon system. Across the cluster, kube-proxy shed its cadvisor dependency [3][4] in favor of cpuset.NumCPU() from k8s.io/utils, a cleaner read of actual node topology that matters for conntrack sizing under static CPU policies. The scheduler gained resourceVersion awareness in Pod status patches [5], tightening consistency guarantees. Test infrastructure is marching forward: CAPZ release-1.25 branch replaces EOL release-1.23 in periodic jobs [6], and the 5k-node scale job got realigned with non-experimental baselines [7]. On the warning front, kubelet now only surfaces missing pull secret alerts when image pulls actually fail [8], cutting noise for clusters that don't rely on private registries.

Action items

References

  1. [1] Add managed Karpenter EC2NodeClass and NodePool ↗ kubernetes/kops
  2. [2] Allow setting missing slice elements from the command line ↗ kubernetes/kops
  3. [3] kube-proxy: detect host CPU count via cpuset.NumCPU instead of cadvisor kubernetes/kubernetes
  4. [4] kube-proxy: detect host CPU count via cpuset.NumCPU instead of cadvisor ↗ kubernetes/kubernetes
  5. [5] scheduler: Add resourceVersion to Pod status patch ↗ kubernetes/kubernetes
  6. [6] Add CAPZ release v1.25 tests ↗ kubernetes/test-infra
  7. [7] Merge pull request #37339 from Jefftree/drop-etcd-version-pull-5k kubernetes/test-infra
  8. [8] Only warn about missing pull secrets if image pull fails ↗ kubernetes/kubernetes

FAQ

What changed in Kubernetes on June 27, 2026?
Kops now ships native Karpenter EC2NodeClass and NodePool support while kube-proxy eliminates its cadvisor dependency by reading CPU topology directly from cpuset.
What should Kubernetes teams do about it?
Upgrade kops to pick up Karpenter integration if you're migrating from external Karpenter configuration • Update any clusters running CAPZ to track release-1.25 branch in periodic test jobs • Monitor cadvisor removal from kube-proxy rollout, especially on CPU-constrained nodes
Which Kubernetes repositories shipped on June 27, 2026?
kubernetes/kops, kubernetes/kubernetes, kubernetes/test-infra

Related across the cluster

For your repos

The showcase is a teaser.
Your wire is the product.

Same engine. Different stack. Below: what changes when the wire is yours.

Showcase wire

  • •14 famous open source orgs
  • •One wire per day
  • •Public, generic
  • •Read on the web, when you remember

Your wire

  • →Up to 1,500 of your repos - orgs, deps, vendors
  • →Morning and evening briefs
  • →Action items routed to your team
  • →Slack delivery, email, breaking-news CVE alerts

Want a hands-on demo first? Ask a current user for an invite link.