Skip to content

FailedScheduling

No node passed the scheduler’s filters, so the pod stays Pending. The event message lists every reason, grouped by how many nodes each one ruled out.

Each clause accounts for some of the nodes. In the example, 1 + 2 + 3 = 6: one control plane node is tainted, two workers are short of memory, three don’t match the selector. The preemption: part says whether evicting lower-priority pods would have helped.

Clause Meaning Check
Insufficient cpu / Insufficient memory Sum of requests on the node plus this pod exceeds allocatable kubectl describe node <node>, section Allocated resources
had untolerated taint(s) The node is tainted and the pod has no matching toleration. Current releases don’t name the taint; older ones printed it in braces kubectl get nodes -o custom-columns=NAME:.metadata.name,TAINTS:.spec.taints
didn't match Pod's node affinity/selector nodeSelector or affinity labels don’t exist on those nodes kubectl get nodes --show-labels
didn't match pod anti-affinity rules Anti-affinity forbids placing it next to existing pods The pod’s affinity.podAntiAffinity
didn't satisfy existing pods anti-affinity rules Pods already on those nodes have anti-affinity rules that exclude this pod The other pods’ affinity.podAntiAffinity
didn't match PersistentVolume's node affinity Its volume is pinned to a zone or node with no usable node (older releases: had volume node affinity conflict) PersistentVolume node affinity
node(s) were unschedulable Nodes are cordoned kubectl get nodes shows SchedulingDisabled
Too many pods The node reached its max pods limit Node allocatable.pods

Fix the specific clause: right-size requests, add capacity, add the toleration or label, or relax the affinity rule. If there are no events at all and the pod stays Pending, the scheduler itself may not be running. See Control plane failures.

Read next: How a pod gets from YAML to running.