WhatBrokeToday
GTA 6FAQ
Kubernetes

Kubernetes EndpointSlice Auto topology hints stay stale after CPU changes

medium

Is it fixed? Not yet.

Ongoing since Nov 17, 17:22 UTC. First reported Nov 17, 17:22 UTC.

What happened

For Services using service.kubernetes.io/topology-mode: Auto, CPU-only Node updates fail to refresh cached zone capacity ratios, leaving topology hints stale and failing to properly recover after fallback.

What caused it?

CPU-only Node updates do not refresh the cached zone capacity ratios for services using automatic topology hints.

Source: github.com

No verified workaround yet

None of the linked reports describe a mitigation. This page updates when one appears.

Timeline

  1. Investigating(current)
  2. Workaround available(upcoming)
  3. Resolved(upcoming)
  1. First report on GitHub issue

    EndpointSlice Auto topology hints can stay stale after CPU changes and fail to recover after fallback

    Source

Evidence

1 report · 1 GitHubShow
  1. [1]
    GitHub First report
    EndpointSlice Auto topology hints can stay stale after CPU changes and fail to recover after fallback(opens GitHub issue in a new tab)

    What happened? For Services using service.kubernetes.io/topology-mode: Auto, CPU-only Node updates do not refresh the cached zone capacity ratios. With two Ready Nodes in different zones, 4 allocatable CPUs per Node, and six Ready endpoints split evenly between them, hints initially allocate 3:3. Updating one Node to 8 allocatable CPUs leaves hints at 3:3 even after the informer observes the update. Recalculating with the existing allocator produces 4:2. There is also a recovery gap: failed…

    by ryo-whaletech 0upvotes 2comments

FAQ

What happened in the Kubernetes issue?

For Services using service.kubernetes.io/topology-mode: Auto, CPU-only Node updates fail to refresh cached zone capacity ratios, leaving topology hints stale and failing to properly recover after fallback.

What is affected by the Kubernetes issue?

Kubernetes.

When did the Kubernetes issue start and end?

It started Nov 17, 17:22 UTC and is still ongoing (1 d 11 h so far). This page updates as it develops.

How do I get alerts for Kubernetes?

Subscribe to the Kubernetes RSS feed for every new Kubernetes incident, or connect an AI assistant to WhatBrokeToday over MCP and ask whether Kubernetes is down. Kubernetes feed

Where does this information come from?

WhatBrokeToday collects public status pages, developer community reports and security advisories, groups the ones about the same problem, and links every report under Evidence on this page. This one has 1 report from 1 independent source.