liveness Probes

Periodically test an app in a container.
If the test fails, the container is destroyed & recreated.
This allows devs to "decide" what makes an app healthy.
NOTE: the livenessProve goes IN the container definition. This applies to a pod AND this can apply in a deployment of pods!

apiVersion: v1
kind: Pod
metadata:
  name: a-dummy-api
  labels:
    name: horse
spec:
  containers:
  - name: api-box
    image: api-box-img
    ports: 
      - containerPort: 8080
    # THIS!
    livenessProbe:
      httpGet:
        path: /api/its-alive
        port: 8080

Other options:

# tcp tests
livenessProbe:
  tcpSocket:
    port: 3306

# bash command tests
livenessProbe:
  exec:
    command: ["cat", "check/if/its/alive"]
  
  # these are available on all livenessProve types
  initialDelaySeconds: 5
  periodSeconds: 1
  failureThreshold: 10

Liveness Probes in a Deployment

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    app: cowdep
  name: cowdep
spec:
  replicas: 1
  selector:
    matchLabels:
      app: cowdep
  strategy: {}
  template:
    metadata:
      creationTimestamp: null
      labels:
        app: cowdep
    spec:
      containers:
      - image: docker/cowsay
        name: cowsay
        command:
        - sh
        - -c
        - "cowsay horselisten"
        # THIS is the liveness probe, IN the container definition block
        livenessProbe:
          exec:
            command: ["cat", "check/if/its/alive"]

Tags:

Kubernetes