Probe

Live Cycle

Pod Status:

• pending: k8s is finding a suitable node to schedulet the pod
• containerCreating: the pod has been scheduled on a node and it is creating container(s).
• Running
• Terminate

Conditions:

• PodScheduled
• Initialized
• ContainersReady
• Ready: is ready to accept request

Readiness Probes

• Ready:
  • Default: Once a container is created, it is ready to accept traffic
  • Custom: Use readiness probes to check whether the application is ready or not.
    • Type:
      • httpGet
      • tcpSocket
      • command
    • More settings:
      • initialDelaySecondes
      • periodsSeconds
      • failureThreshold

pod.yaml

apiVersion: v1
kind: Pod
metadata:
  name: myweb
  labels:
    name: myweb

spec
  containers:
    - name: mycontainer
      image: mycontainer
      ports:
        - containerPort: 8080
      readinessProbe:
        httpGet:
          path: /index.html
          port: 8080

---

Liveness Probes

• If a pod is crashed, k8s will keep restart it. However, if a web application within the pod is crashed, we need to use liveness probes to check and create a new pod.
• The type and more settings is the same as readinessProbe.

pod.yaml

apiVersion: v1
kind: Pod
metadata:
  name: myweb
  labels:
    name: myweb

spec
  containers:
    - name: mycontainer
      image: mycontainer
      ports:
        - containerPort: 8080
      livenessProbe:
        httpGet:
          path: /index.html
          port: 8080