A toleration is a pod scheduling specification that permits placement on a node with a matching taint. Taints repel pods unless their specifications include the required toleration; a toleration permits scheduling but does not select a particular node. section 2.1.3
Use tolerations when an operational task requires a pod to run on a tainted node:
- Match the pod specification to the taint’s key, value when present, and effect; a taint requires a key and effect but may omit a value. section 2.1.3
Allowing a pod onto a tainted node
A control-plane node commonly carries the taint node-role.kubernetes.io/control-plane:NoSchedule, which prevents ordinary application pods from being scheduled there. CoreDNS provides a concrete comparison: inspect its tolerations with kubectl get pod ... -o yaml -n kube-system | grep tolerations -A14, then compare the result with the node taint. For a hands-on check, generate pod-tolerate.yaml, add the required toleration, create it, and verify placement with kubectl get pods -o wide. section 2.1.3
