Node Selector

For multiple reasons, there can be a need to run a pod on a specific worker node.

Node affinity is a set of rules used by the scheduler to determine where a pod can be placed.

In Kubernetes terms, it is known as nodeSelector, and nodeAffinity/podAffinity fields under PodSpec.

In Kubernetes, we can achieve nodeAffinity with the help of:

nodeSelector

nodeAffinity (more flexibility)

Node affinity is conceptually similar to nodeSelector – it allows you to constrain which nodes your pod is eligible to be scheduled on, based on labels on the node.

We can label our nodes using key-value pairs and use these labels to schedule a pod on a given node. Node selectors have limitations. We can only use a single label and selector to select a pod. We cannot use complex logic to decide which node should be used for scheduling a pod. Example: we cannot specify a pod to be scheduled on a node with compute: high or compute:medium at the same time. For such use cases, we need

Node Affinity.

Select a node for a pod

Example: schedule the pod on a node with high compute power.

kubectl label node node01 compute=high


apiVersion: v1
kind: Pod
metadata:
	name: web-pod
spec:
	nodeSelector:
		compute: high
	containers:
		- name: nginx
			image: nginx

Commands

Label a node - kubectl label node <node-name> <key>=<value>

View labels of a node - k describe node <node-name>