How to Update a K8s Deployment with More Replicas

Question

I've created a simple K8s deployment with the kubectl create command

kubectl create -f k8-deployment.yaml

My k8-deployment.yaml file looks like this

apiVersion: apps/v1
kind: Deployment
metadata:
  creationTimestamp: null
  labels:
    app: mage-di
  name: mage-di
spec:
  replicas: 1
  selector:
    matchLabels:
      app: mage-di
  strategy: {}
  template:
    metadata:
      creationTimestamp: null
      labels:
        app: mage-di
    spec:
      containers:
      - image: astorm/mage-di
        name: mage-di
        imagePullPolicy: Never
        resources: {}
status: {}

This results in a single pod being started.

I want to tell my k8 cluster that more pods are needed to handle an expected traffic spike.

How should I do this? If I look at kubectl help I see there's an edit command that allows me to edit a deployment object's configuration, but this requires an interactive editor. Also, I'm new to K8s and I'm unsure if editing a deployment in place and updating its replica count is enough to trigger the proper creation of new pods. If I look at other kubectl commands I see there's also a rollout, apply and patch command that might do what I want.

Is there a canonically accepted way to do what I want, or is K8s the sort of tech where I just need to experiment and hope for the best?

Jeffiejeffrey

If you have some way of measuring how much load you're handling, you can also look at setting up the Horizontal Pod Autoscaler to set the replicas: field automatically. – Levee

Answer 1

You can do this in two ways. Either imperative - a quick command Or declarative - good for a production environment where you store your Deployment-manifest in Git.

Imperative way: (this will then diverge from what you have in your yaml-file)

kubectl scale deployment mage-di --replicas=2

Declarative way, edit this line in your Yaml file:

replicas: 2

then apply it to the cluster with:

kubectl apply -f k8-deployment.yaml

See also:

• Declarative config management
• Imperative config management with commands

Spark