Who contributed to Kubernetes on August 22, 2026?

13 developers shipped this update, including Peter Rifel, kubernetes-prow[bot], hakman, William Yao, SwathiR03, karpovantonme, my-git9, and yysyo111, and 5 more.

What were the notable Kubernetes updates?

kops: drop the kindnet grid jobs for kops 1.34, kops: drop the kindnet grid jobs for kops 1.34, and kops: raise the conformance timeout to 200m and quiet the test nodes.

@kubernetes

Container orchestration - what platform teams ship on

github ↗

Pick a date

Topics: Go Infrastructure Full archive →

The Wire · Showcase

KOPS DROPS KINDNET GRID JOBS AMID DNS FAILURES

By RepoJournal · Filed 06:56 UTC on August 22, 2026 · About Kubernetes

13 people shipped this

Peter Rifel @rifelpet 4 cited

hakman @hakman 1 cited

SwathiR03 @SwathiR03 1 cited

tico88612 @tico88612 1 cited

kubernetes-prow[bot] @kubernetes-prow[bot]

William Yao @willie-yao

karpovantonme @karpovantonme

my-git9 @my-git9

yysyo111 @yysyo111

dependabot @dependabot

Nir Soffer @nirs

minikube-bot @minikube-bot

jrvaldes @jrvaldes

The kops test grid is shedding 27 kindnet periodics after a node-local DNS cache started failing live Services, and conformance timeouts are getting a much-needed bump.

kops' kindnet grid jobs for version 1.34 are being dropped entirely [1], a move that eliminates 27 periodics (12 AWS, 15 GCE) after the node-local DNS cache intermittently answered NXDOMAIN for live Services, failing a [sig-network] spec with "service is not reachable within 2m0s timeout" [2]. The cache, enabled by kops' own manifest flags, has been defaulted off upstream since kindnet 629dc20ec6 [1]. Meanwhile, the kops conformance suite no longer fits in its 150m timeout, causing the Go test binary to panic and produce no junit at all, so the team is raising the timeout to 200m and quieting test nodes [3]. On the tooling front, a new dump capture for Calico BGP session state aims to leave evidence behind for the GCE Calico grid failures, addressing the dead end where artifacts only showed symptoms [4]. Also, a live race in nodeup is being fixed: systemd-networkd restarts can land mid-apt-get, and the new ordering prevents that [5]. In other news, the v1.37 Release Announcement Blog is in progress [6], and KEP-5207 has been marked as implemented [7].

Action items

→ Review kops 1.34 grid changes if you rely on kindnet periodics kubernetes/test-infra [monitor]
→ Upgrade kops conformance jobs to 200m timeout to avoid blank testgrid kubernetes/test-infra [plan]
→ Test the Calico BGP dump capture in kops toolbox dump kubernetes/kops [plan]

References

[1] kops: drop the kindnet grid jobs for kops 1.34 kubernetes/test-infra
[2] kops: drop the kindnet grid jobs for kops 1.34 ↗ kubernetes/test-infra
[3] kops: raise the conformance timeout to 200m and quiet the test nodes ↗ kubernetes/test-infra
[4] dump: capture Calico BGP session state from each node ↗ kubernetes/kops
[5] nodeup: order systemd-networkd restarts after package installs ↗ kubernetes/kops
[6] Kubernetes v1.37 Release Announcement Blog ↗ kubernetes/website
[7] KEP-5207: marked as implemented ↗ kubernetes/enhancements

Quick answers

What shipped in Kubernetes on August 22, 2026?: The kops test grid is shedding 27 kindnet periodics after a node-local DNS cache started failing live Services, and conformance timeouts are getting a much-needed bump. In total, 39 commits and 27 pull requests landed.
Who contributed to Kubernetes on August 22, 2026?: 13 developers shipped this update, including Peter Rifel, kubernetes-prow[bot], hakman, William Yao, SwathiR03, karpovantonme, my-git9, and yysyo111, and 5 more.
What were the notable Kubernetes updates?: kops: drop the kindnet grid jobs for kops 1.34, kops: drop the kindnet grid jobs for kops 1.34, and kops: raise the conformance timeout to 200m and quiet the test nodes.

Go 1.26.5 image gap stalls Kubernetes patch builds

A missing Go patch release is bricking patch builds for Kubernetes 1.34 and 1.35, and the fix just merged.

go 24 shipped

@kubernetes Aug 20

Kubernetes ships v1.37.0-rc.1

Kubernetes drops its first release candidate for v1.37, and a scheduler test flake gets fixed.

go 4 shipped

@kubernetes Aug 19

CI Flake Hunters Strike Gold in GCP Test Harness

Kubernetes CI's most flaky test just got a surgical fix, and it's the kind of cleanup that keeps your pipelines honest.

go 67 shipped

@kubernetes Aug 18

KOPS BRINGS AKAMAI ETCd MANAGEMENT TO THE FOREFRONT

Kops is about to make etcd volume management on Akamai a first-class citizen, and the CI pipeline is getting smarter about SSH user discovery.

go 72 shipped

Elsewhere on the wire

AI Agents Aug 22

Cline hubs get drain and replay as Langchain plumb gates

Cline just shipped rollover-grade hub commands while Langchain exposes gateway internals, and OpenHands quietly broke its own monitoring contract.

ai-agents 42 shipped

Local LLMs Aug 22

SGLang ships v0.5.18 with 710 PRs and DeepSeek V4 FP4 auto-routing

The biggest release of the week just landed: SGLang v0.5.18 adds new models, fixes a silent disagg bug, and defaults DeepSeek V4 FP4 checkpoints to the right MoE runner.