2.04 Node Selectors

• Cluster has 3 nodes: 2 small (limited resources) and 1 large (higher capacity).
• Heavy data-processing jobs should run on the large node.
• By default, Kubernetes can schedule pods on any node.
• This may place heavy pods on small nodes, which is not desired.
• To avoid this, restrict pods to specific nodes using node selectors (simplest method).

---

NodeSelector Example

Important

• Nodes are labeled with key-value pairs (e.g., size=large).
• The scheduler uses these labels to decide where to place pods.
• Nodes must be labeled before using them in a nodeSelector.

Node Labeling Command

kubectl label nodes <node-name> <label-key>=<label-value>
# Example: kubectl label nodes node-1 size=Large

Pod Configuration with NodeSelector (Click to Expand)

apiVersion: v1
kind: Pod
metadata:
  name: myapp-pod
spec:
  containers:
    - name: data-processor
      image: data-processor
  nodeSelector:
    size: Large

• In the pod definition file, you can add a nodeSelector under the spec section with size: large to ensure the data-processing pod runs only on the larger node.

• This works because nodes are assigned labels (e.g., size=large), and the Kubernetes scheduler matches these labels with the node selector to decide where the pod should run.

---

Node Selector Limitations

• NodeSelector works for simple cases with a single label.
• Complex rules (e.g., run on large OR medium nodes, or avoid small nodes) are not possible with NodeSelector.
• For such scenarios, use Node Affinity and Anti-Affinity.