#NoSchedule
4. Is the NODE the real patient?

kubectl describe node <node> | grep -A8 Conditions:

DiskPressure True adds a NoSchedule taint, so new pods stop landing there. That's why a wave of Evicted is followed by a wave of Pending.

One node = clean it. Fleet = capacity.
October 21, 2026 at 6:08 AM
Kubernetes taints and tolerations let nodes push pods away by default. Learn when to use them, how NoExecute differs from NoSchedule, and why this matters for…

https://dev.to/siddharthajmore/when-nodes-start-rejecting-pods-taints-and-tolerations-1kf3

#DevOps #platform
September 23, 2026 at 3:30 PM
RETIREMENT: WHERE EVERY DAY IS SATURDAY, BUT YOU STILL FORGET WHICH ONE.
#RetiredLife #GrumpyOldMan #NoSchedule
September 8, 2026 at 10:05 PM
Isolate GPUs in K8s: apply a NoSchedule taint (hardware=nvidia-a100) so kube-scheduler rejects any pod lacking a matching toleration during node filtering. #kubernetes #taints #tolerations

https://www.valtersit.com/vault/dedicated-gpu-isolation-with-noschedule-taint-and-pod-tolera-627321/
August 24, 2026 at 1:21 AM
The first suggestion from this answer is what fixed it: unix.stackexchange.com/a/791920, but not before I migrated from Cilium to Flannel after mistakenly believing this was a Cilium problem (Cilium was failing to start and had labelled my nodes NoSchedule).
October 14, 2025 at 7:47 AM
GKE now supports opting out of the default kubernetes.io/arch=arm64:NoSchedule taint on Arm nodes in Standard node pools and in custom ComputeClasses
July 28, 2026 at 9:20 PM
GKE now supports opting out of the default kubernetes.io/arch=arm64:NoSchedule taint on Arm nodes in Standard node pools and in custom ComputeClasses
July 28, 2026 at 9:20 PM