176 wires and counting

$ follow Kubernetes

Keep up with Kubernetes in about 3 minutes: what actually shipped — the commits, pull requests, releases, and security advisories that matter.

or

fair warning: these emails are deeply technical. diffs, version numbers, CVEs, benchmark deltas. if that's not your idea of a good read, this isn't your newsletter.

Folds into your digest — weekly by default, monthly if you prefer. Unsubscribe in one click.

$ status

wire 2026-10-04
stories 29

© 2026 RepoJournal Home Showcase How it works Privacy

$ the-wire · showcase

Per-VPA target percentiles, WatchList benchmarks in the request path

By RepoJournal · Filed · About Kubernetes · Composed from the cited sources · methodology

The day's work spans VPA configuration surface and the measurement plumbing for WatchList, the API mechanism the project is benchmarking before it can replace the watch cache.

AEP-9970: per-VPA target percentiles kubernetes/autoscaler

by Sanil2108

The AEP adds targetCPUPercentile and targetMemoryPercentile to ContainerResourcePolicy, so the percentile behind a VPA's CPU and memory recommendations becomes a per-object setting rather than a global flag. It extends AEP-8026's per-VPA configuration mechanism to the target fields; bound percentiles are explicitly left as future work.

add watch-list mode to request benchmark kubernetes/perf-tests

by p0lyn0mial

The request benchmark gains an optional HTTP WatchList mode, which measures WatchList latency while ruling out client-side typed-object decoding as a source of overhead. The companion presubmit job creates pull-kubernetes-benchmark-watchlist-proto-http to exercise the WatchList HTTP list path in CI.

Migrate 5,000-node scale tests to c4d-standard-96 kubernetes/test-infra

by Jefftree

All 5,000-node scale test control planes move from c4-standard-96 to c4d-standard-96 after the upstream jobs showed severe run-to-run variance from host processor differences in the provider pool, which was obscuring genuine performance regressions. The experiment is done and the backing revert is included.

Merge pull request #142638 from Jefftree/event-handler-name-portability kubernetes/client-go

by Kubernetes Publisher

TestNameForHandler in client-go/tools/cache now derives the package path dynamically instead of hardcoding it. Small, but it removes a portability break for anyone running that test outside its original tree.

Merge pull request #10079 from locker95/docs/vpa-requests-limits-faq kubernetes/autoscaler

by kubernetes-prow[bot]

Elsewhere: the autoscaler FAQ now documents requests-versus-limits behavior, DRANET 1.5.0 images are promoted to the prod registry, and node-ipam-controller switches to squash merges.

Quick answers

What shipped in Kubernetes on October 4, 2026?
The day's work spans VPA configuration surface and the measurement plumbing for WatchList, the API mechanism the project is benchmarking before it can replace the watch cache. In total, 19 commits and 10 pull requests landed.
Who contributed to Kubernetes on October 4, 2026?
7 developers shipped this update, including Sanil2108, kubernetes-prow[bot], p0lyn0mial, Jefftree, supahero1, Gaurav Ghildiyal, and Kubernetes Publisher.
What were the notable Kubernetes updates?
AEP-9970: per-VPA target percentiles, add watch-list mode to request benchmark, and Migrate 5,000-node scale tests to c4d-standard-96.