Cloud-Native Microservices With Kubernetes - 2nd Edition

The nodeSelector: Simple Custom Scheduling with Labels

This is the easiest way to constrain Pods to specific nodes, as it uses a simple label selector. It works according to a simple principle:

• You label your nodes with key-value pairs.
• You specify those key-value pairs in your Pod specification using the nodeSelector field.
• Kubernetes will then schedule the Pod only on nodes that have matching labels.

nodeSelector: A Practical Example

To better understand how nodeSelector works, let's take a look at a practical example.

Imagine this scenario: You have an Nginx server running a CDN service, and because it serves static content and has high I/O operations, you want it to run only on nodes with SSD disks. In your cluster, you have nodes with both HDD and SSD disks.

The initial setup could look like this:

kubectl apply -f - <
apiVersion: apps/v1
kind: Deployment
metadata:
  name: cdn
spec:
  replicas: 3
  selector:
    matchLabels:
      app: cdn
  template:
    metadata:
      labels:
        app: cdn
    spec:
      containers:
      - name: cdn
        image: nginx
        ports:
        - containerPort: 80
EOF

To schedule the Nginx Pods only on nodes with SSD disks, you first need to identify the nodes with SSD disks and label them accordingly. For this example, let's assume you label those nodes with disktype=ssd. We have two nodes in this example; the first is SSD-based, while the second is HDD-based.

# Any random node will do for this example
export ssd_node=$(kubectl get nodes -o custom-columns=NAME:.metadata.name --no-headers | head -n 1)

# Label the node
kubectl label node $ssd_node disktype=ssd

# Verify the label was added
kubectl get nodes --show-labels | grep $ssd_node

Alternatively, you can transform the output to show a more readable format:

kubectl get node $ssd_node --show-labels=false \
  -o jsonpath='{.metadata.labels}' | jq

You should see a group of key-value pairs representing the labels of the node:

{
  "beta.kubernetes.io/arch": "amd64",
  "beta.kubernetes.io/instance-type": "s-2vcpu-4gb",
  "beta.kubernetes.io/os": "linux",
  "disktype": "ssd",
  "doks.digitalocean.com/managed": "true",
  "doks.digitalocean.com/node-id"