Is Google Kubernetes Engine down?
Last 90 days
99.734% uptime · 2 incidentsWhat Google Kubernetes Engine does - and how it fails
Google Kubernetes Engine runs the managed control plane your clusters depend on - the API server, scheduler, and etcd. The critical nuance: the GKE SLA covers the control plane, not your workloads. Well-built clusters keep serving traffic through control-plane incidents; what breaks is deploys, autoscaling, and anything that calls the Kubernetes API on its hot path.
- Kubernetes API errors or timeouts (kubectl hangs, CI/CD deploys fail) while pods keep running
- Stuck cluster operations: node-pool upgrades, cluster creation, or repairs not progressing
- Autoscaler paralysis - HPA/cluster-autoscaler can’t act because the control plane is degraded
- Node-pool level problems in one zone (node registration failures, workload rescheduling storms)
Recent Google Kubernetes Engine incidents
Multiple products in us-central1-b are experiencing network service degradation.
September 1, 2026 · 4h 8mMultiple products in us-central1-b are experiencing network service degradation.
We are investigating an issue where customers may experience timeouts, service degradations, errors, and elevated latencies across multiple products in the us-west1 region.
August 20, 2026 · 3h 40mWe are investigating an issue where customers may experience timeouts, service degradations, errors, and elevated latencies across multiple products in the us-west1 region.
Full analyses live in the post-mortem archive.
Google Kubernetes Engine questions
Is GKE down right now?
The live pill above reflects our latest check. GKE incidents are usually control-plane scoped: if kubectl fails but your Services still answer traffic, that pattern is exactly a control-plane event - your data plane is fine.
My pods are running but kubectl times out - is that a GKE outage?
Very likely, yes - and it is the most common GKE failure mode. The control plane (API server) is a Google-managed regional or zonal component. Workloads on the nodes continue serving; deploys, scaling actions, and operators that watch the API are what stall.
What does the GKE SLA actually cover?
A regional cluster’s control plane carries a 99.95% monthly uptime SLA (zonal control planes are 99.5%). It covers Kubernetes API availability - not your pods, nodes, or application traffic. Workload availability is your architecture’s job: multi-zone node pools, PodDisruptionBudgets, and not depending on the API server at request time.
Get alerted when Google Kubernetes Engine breaks
Instant email the moment we detect an incident - affected services and regions included.
Get outage alerts