Edit on GitHub

Deployment

• Manage pods lifecycles
• Ensure high availability
• Enable scaling

Quick deploy and expose

Create a deployment with a service and a pod, and then expose the pod on the node

kubectl create deployment nginx-skobba --image=nginx --port=80

kubectl expose pod nginx-skobba-xxxxxxx --type NodePort --port 80

# or

kubectl expose deployment nginx-skobba --type=NodePort --port=80

Sample with Volumes

First create pv and pvc.

cat <<EOF | kubectl apply -f -
apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: host-path-deploy
  name: host-path-deploy
  namespace: skobba
spec:
  replicas: 1
  selector:
    matchLabels:
      app: host-path-deploy
  template:
    metadata:
      labels:
        app: host-path-deploy
    spec:
      containers:
      - image: nginx:alpine
        name: nginx
        volumeMounts:
        - name: host-path-volume
          mountPath: /myvol
      volumes:
      - name: host-path-volume
        persistentVolumeClaim:
          claimName: host-path-pvc
EOF

History

# Show history
kubectl rollout history deployment mydep

# Show a spesific revision
kubectl rollout history deployment mydep --revision=3

# Show status
kubectl rollout status deployment mydep

Rollback

# Rollback one (not the new rev number)
kubectl rollout undo deployment mydep

# Rollback to a spesific version
kubectl rollout undo deployment mydep --to-revision=3

Convert pod into deployment

1. Export the pod

kubectl get pod -n myns mypod -oyaml

2. Manually Remove Unnecessary Fields:

metadata.creationTimestamp
metadata.resourceVersion
metadata.selfLink
metadata.uid
metadata.annotations (if it contains auto-generated data like kubectl.kubernetes.io/last-applied-configuration)
status (the entire status section can be removed)
spec.nodeName
spec.serviceAccountName.

3. Change kind from Pod to Deployment:

kind: Deployment

4. Add the apiVersion for Deployments:

apiVersion: apps/v1
kind: Deployment

5. Wrap the spec in a template and add selector:

• Under the spec field, add a template field, and move the entire original pod spec under it.
• Template field should contain the pod’s metadata and spec.

apiVersion: apps/v1
kind: Deployment
metadata:
  name: mydep
spec:
  selector:
    matchLabels:
      app: myapp
  template:
    metadata:
      labels:
        app: myapp
    spec:
      containers:
      - name: mycontainer
        image: nginx
        ports:
        - containerPort: 80

6. Add a replicas field to define how many replicas you want:

• Set this under the spec field at the same level as selector.

apiVersion: apps/v1
kind: Deployment
metadata:
  name: mydep
spec:
  selector:
    matchLabels:
      app: myapp
  template:
    metadata:
      labels:
        app: myapp
    spec:
      containers:
      - name: mycontainer
        image: nginx
        ports:
        - containerPort: 80

Sample

Pod:

apiVersion: v1
kind: Pod
metadata:
  name: mypod
spec:
  containers:
  - name: mycontainer
    image: nginx
    ports:
    - containerPort: 80

Deployment:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: mydep
spec:
  replicas: 3
  selector:
    matchLabels:
      app: myapp
  template:
    metadata:
      labels:
        app: myapp
    spec:
      containers:
      - name: mycontainer
        image: nginx
        ports:
        - containerPort: 80

Check Resource Limits

kubectl get deployments --all-namespaces -o custom-columns=NAMESPACE:.metadata.namespace,NAME:.metadata.name,CPU_REQUEST:.spec.template.spec.containers[*].resources.requests.cpu,MEM_REQUEST:.spec.template.spec.containers[*].resources.requests.memory,CPU_LIMIT:.spec.template.spec.containers[*].resources.limits.cpu,MEM_LIMIT:.spec.template.spec.containers[*].resources.limits.memory