Labels & Annotations

Add label to a Node

oc label node NODE_NAME node-role.kubernetes.io/storage=

Above command is used to mark a node as a storage node using the label “node-role.kubernetes.io/storage=”

Add node selector at namespace level

In order to run the pods of a namespace on a specific node, we add can a node selector at namespace level

For example, If a node has a label “node-role.kubernetes.io/worker=”

oc annotate namespace NAMESPACE 'openshift.io/node-selector=node-role.kubernetes.io/worker='

Add taints on a node

To control the scheduling of pods on a node, we can add taints to a node.

There are different types of default Taints available. For example, the effect “NoSchedule” prevents the scheduling of pods on a node which has this taint

oc adm taint nodes NODE_NAME KEY=VALUE:NoSchedule

oc adm taint nodes NODE_NAME node-role.kubernetes.io/storage:NoSchedule

Above command prevents the scheduling of pods on the nodes using the key name “node-role.kubernetes.io/storage” and the Operator “Exists” (as we have not defined any value here). To allow pods to run on the tainted node, one must add tolerations in the pod.

    tolerations:
        - key: node-role.kubernetes.io/storage
          operator: Exists
          effect: NoSchedule

Add tolerations at namespace level

oc annotate namespace NAMESPACE 'scheduler.alpha.kubernetes.io/defaultTolerations=[{"operator": "Exists", "effect": "NoSchedule", "key": "node-role.kubernetes.io/storage"}]'

In the above command, we added the toleration for the taint “NoSchedule” having key “node-role.kubernetes.io/storage” and the operator “Exists”