Skip to content
OpenAI ยท AIDec 11, 2024, 23:16 UTC

New telemetry service overwhelms Kubernetes control planes

CriticalDeploymentUpdated 16h ago
Dec 11, 23:16 UTCDec 12, 03:38 UTC
Duration
4h 22m
Impact
Critical
Root cause
Deployment
OpenAI, 90 days
26 incidents
Affected
ChatGPTAPISoraGlobal

Lesson: Test changes at production cluster size, and keep break-glass access to the control plane that does not depend on the thing that is failing.

What happened

A new telemetry service made every node run expensive Kubernetes API operations, overwhelming the API servers in the largest clusters. DNS-based service discovery depends on the control plane, so services failed as DNS caches expired, and engineers were locked out of the control plane needed to roll back.

More from OpenAI

Full history

Also caused by deployment or rollout

All
StartedIncidentDuration
Sep 1106:22 UTCIssues with Workers VPC hostname route resolution on 2026-09-11Cloudflare0m
Aug 2014:43 UTCIntermittent failures creating agent tasksGitHub9h 54m
Aug 2000:31 UTC[Medium] Issue with Microsoft Office Integration and Box EditBox1h 13m
Aug 622:22 UTCTrouble Using Search Bar For Some AdminsSlack2h 44m
Jul 2110:23 UTCJob runs failing at the git clone stepdbt Labs2h 29m
Jul 1907:33 UTCSome EMEA customers experiencing Git clone failures (403 errors) on thier accountsdbt Labs6h 22m

From vendors' own status pages and disclosures. Times as reported. Logos via logo.dev; trademarks belong to their owners.

Weekly: the week's major outages, postmortems and breaches, Saturday mornings.