nodeSelector

NodeSelector is a field in the Kubernetes Pod specification that constrains the scheduler to place a Pod only on nodes that have matching labels. It provides a simple mechanism to ensure Pods run on nodes with certain characteristics, such as hardware type, geographic location, or policy requirements.

How it works: The scheduler compares the nodeSelector map against the labels on each candidate Node. A node is eligible only if it has each key in nodeSelector with the corresponding value. If no eligible node exists, the Pod remains in Pending state until a matching node becomes available. NodeSelector is a basic, equality-based constraint and does not express preferred or taint-based rules.

Limitations and relation to other features: For more expressive scheduling, Kubernetes offers nodeAffinity and taints/tolerations. NodeSelector cannot express soft preferences or complex requirements, and it is evaluated at scheduling time; changes to node labels after a Pod is scheduled do not automatically re-schedule it. If you need to move a running Pod, you typically delete and recreate it (or update the controller) so that new pods are scheduled with the updated constraints.

Example:

apiVersion: v1

kind: Pod

metadata:

name: example

spec:

nodeSelector:

disktype: ssd

containers:

- name: app

image: myapp:latest

Practical notes: To use nodeSelector, label nodes accordingly (for example, kubectl label nodes <node-name> disktype=ssd). Pods created by controllers such as Deployments will inherit the same scheduling constraints for their new pods.