Organize resources in a namespace

Virtual cluster inside a cluster

Default namespaces

• kube-system, system processes
• kube-public, publicely accessible data
• kube-node-lease, heartbeats of nodes
• default, default namespace

# create a namespace
# kubectl create namespace [namespaceName]
kubectl create namespace count-ns

# create a resource in a namespace
kubectl apply -f redis-deployment.yaml --namespace=count-ns
        

# namespace.yaml
apiVersion: v1
kind: Namespace
metadata:
  name: count-ns
        

# list namespace
kubectl get namespaces

# list resources in a namespace
# kubectl -n [namespaceName] get [resource]
kubectl -n count-ns get all
kubectl -n count-ns get services
        

# install kubens
brew install kubectx
        

# list namespace and highlight the current namespace
# by default, "default" is the current namespace
kubens

# switch namespace
# kubens [namespaceName]
kubens count-ns
kubenctl get all # list resources in current namespace
        

Cannot access most resource from another namespace

Each namespace must define own ConfigMap, Secret

Service can be shared, access a service in a different namespace with [serviceName].[namespaceName]

Some components cannot be created in namespace, such as volume, node

1. Create a namespace for each team or project

# app.yaml
apiVersion: v1
kind: Namespace
metadata:
  name: count-ns

---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: webapp-deployment
  labels:
    app: webapp
  namespace: count-ns
spec:
  replicas: 1
  selector:
    matchLabels:
      app: webapp # let deployment know which pod for this deployment
  template: # configuration for pods
    metadata:
      labels:
        app: webapp # each pod has a unique name, but pods can share the same label
    spec:
      containers: # containers in a pod, usually add one container per pod
      - name: webapp
        image: lchenlangley/count # pull image from local
        imagePullPolicy: Never
        ports:
        - containerPort: 5000
        env:
        - name: MYREDIS_HOST
          value: redis-service

---
apiVersion: v1
kind: Service
metadata:
  name: webapp-service # service name accessed by app
  namespace: count-ns
spec:
  selector:
    app: webapp # match the pod labels
  ports:
    - protocol: TCP
      port: 5000 # service port
      targetPort: 5000 # pod port

---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: redis-deployment
  labels:
    app: redis
  namespace: count-ns
spec:
  replicas: 1
  selector:
    matchLabels:
      app: redis # let deployment know which pod for this deployment
  template: # configuration for pods
    metadata:
      labels:
        app: redis # each pod has a unique name, but pods can share the same label
    spec:
      containers: # containers in a pod, usually add one container per pod
      - name: redisdb
        image: redis:alpine # pull image from Docker Hub
        ports:
        - containerPort: 6379

---
apiVersion: v1
kind: Service
metadata:
  name: redis-service # service name accessed by app
  namespace: count-ns
spec:
  selector:
    app: redis # match the pod labels
  ports:
    - protocol: TCP
      port: 6379 # service port
      targetPort: 6379 # pod port
        

# deploy resources
kubectl apply -f app.yaml

# access app
kubectl -n count-ns port-forward svc/webapp-service 8000:5000
http://localhost:8000/

# remove resources
kubectl delete -f app.yaml
        

2. Resources grouped in namespaces

# monitor.yaml

apiVersion: v1
kind: Namespace
metadata:
  name: monitor-ns

---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: webapp-deployment
  labels:
    app: webapp
  namespace: monitor-ns
spec:
  replicas: 1
  selector:
    matchLabels:
      app: webapp # let deployment know which pod for this deployment
  template: # configuration for pods
    metadata:
      labels:
        app: webapp # each pod has a unique name, but pods can share the same label
    spec:
      containers: # containers in a pod, usually add one container per pod
      - name: webapp
        image: lchenlangley/count # pull image from local
        imagePullPolicy: Never
        ports:
        - containerPort: 5000
        env:
        - name: MYREDIS_HOST
          value: redis-service.db-ns

---
apiVersion: v1
kind: Service
metadata:
  name: webapp-service # service name accessed by app
  namespace: monitor-ns
spec:
  selector:
    app: webapp # match the pod labels
  ports:
    - protocol: TCP
      port: 5000 # service port
      targetPort: 5000 # pod port
        

# db.yaml

apiVersion: v1
kind: Namespace
metadata:
  name: db-ns

---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: redis-deployment
  labels:
    app: redis
  namespace: db-ns
spec:
  replicas: 1
  selector:
    matchLabels:
      app: redis # let deployment know which pod for this deployment
  template: # configuration for pods
    metadata:
      labels:
        app: redis # each pod has a unique name, but pods can share the same label
    spec:
      containers: # containers in a pod, usually add one container per pod
      - name: redisdb
        image: redis:alpine # pull image from Docker Hub
        ports:
        - containerPort: 6379

---
apiVersion: v1
kind: Service
metadata:
  name: redis-service # service name accessed by app
  namespace: db-ns
spec:
  selector:
    app: redis # match the pod labels
  ports:
    - protocol: TCP
      port: 6379 # service port
      targetPort: 6379 # pod port
        

# deploy
kubectl create -f k8s

# access service
kubectl -n monitor-ns port-forward svc/webapp-service 8000:5000
http://localhost:8000/

# delete
kubectl delete -f k8s
        

3. Resources sharing

Deleting the namespace also deletes all the residing components

Removing all deployments within a namespace does not remove the namespace

# kubectl delete namespace [namespaceName]
kubectl delete namespace count-ns
        

Documentation

Count Sample

Command-line tools for Kubernetes: kubectl, stern, kubectx, kubens

Kubernetes best practices: Organizing with Namespaces