LivenessProbe

A pod is considered as ready when all of its containers are ready.
In order to verify for container enough port is healthy and ready to serve traffic, Cuba Netes provides for a range of healthy checking mechanism.
Health checks or prbes are carried out the kubelet to determine when to restart a container (for liveness probe) And use my services and deployments to determine if a prod should receive traffic.

For example - Liveness probes could catch a deadlock, where an application is running but unable to make progress. Restoring a container in such a state can be helpful to make the application more available despite bugs.

Points to Remember

One use of readiness promise to control which part are used as vacants for services. the report is not ready, it is removed from service load balancer.
For running health checks, we would use CMD’s specific to the application.
If the CMD succeeds, it returns zero. And the kubelet considers the container to be alive and healthy if the command return norm zero value the kubelet will restart the pod and recreate it.
In the liveness probe Redefine 3 parameters -

Parameter	Descripiton
`command`	The command whihc will beused fo rhealth check
`intialdelayseconds`	After how much seconds the liveness probe will start health checks when the pod will be created
`periodseconds`	After how many seconds it will keep on checking periodically.
`timeoutseconds`	Time in seconds when the health check keep failing, liveliness probe will declare the pod as dead

Implementation

k8s_example_15.yml


apiVersion: v1
kind: Pod
metadata:
  name: example-15
  labels:
    test: liveness
spec:
  containers:
    - name: container0
      image: ubuntu
      args: 
        - "/bin/sh"
        - "-c"
        - "touch /tmp/healthy; sleep 1000"
      livenessProbe:                                          
        exec:
          command:                                         
            - "cat"
            - "/tmp/healthy"
        initialDelaySeconds: 5          
        periodSeconds: 5                                 
        timeoutSeconds: 30

Create pod and get the details


kubectl apply -f k8s_example_15.yml 
kubectl describe pods

Output


controlplane $ kubectl apply -f file.yml 
pod/example-15 created
controlplane $ kubectl describe pods
Name:             example-15
Namespace:        default
Priority:         0.
.
.
.
    Liveness:       exec [cat /tmp/healthy] delay=5s timeout=30s period=5s #success=1 #failure=3
    Environment:    <none>
    Mounts:
      /var/run/secrets/kubernetes.io/serviceaccount from kube-api-access-ntqtc (ro)
Conditions:
.
.
.
Events:
  Type    Reason     Age   From               Message
  ----    ------     ----  ----               -------
  Normal  Scheduled  23s   default-scheduler  Successfully assigned default/example-15 to node01
  Normal  Pulling    22s   kubelet            Pulling image "ubuntu"
  Normal  Pulled     18s   kubelet            Successfully pulled image "ubuntu" in 3.758s (3.758s including waiting). Image size: 29754422 bytes.
  Normal  Created    18s   kubelet            Created container container0
  Normal  Started    18s   kubelet            Started container container0

Execute the health check command and check for the cmd result, also execute random command for errors to generate non-zero output


kubectl exec -it example-15 -- /bin/bash

Output


controlplane $ kubectl exec -it example-15 -- /bin/bash
root@example-15:/# cat /tmp/healthy
root@example-15:/# echo $?
0
root@example-15:/# cat /tmp/xander 
cat: /tmp/xander: No such file or directory
root@example-15:/# echo $?
1
root@example-15:/# rm -rf /tmp/healthy 
root@example-15:/# cat /tmp/healthy
cat: /tmp/healthy: No such file or directory
root@example-15:/# echo $?
1
root@example-15:/# cat /tmp/brook  
cat: /tmp/brook: No such file or directory
root@example-15:/# echo $?
1
root@example-15:/# command terminated with exit code 137

If you notice once the command failed thrice - we get the prompt saying command terminated with exit code 137 that is when the pod is restarted and new container is created (check the pod description provided)


kubectl describe pods

Output


controlplane $ kubectl describe pods example-15
Name:             example-15
Namespace:        default
Priority:         0
.
.
.
    State:          Running
      Started:      Fri, 18 Oct 2024 18:27:41 +0000
    Last State:     Terminated
      Reason:       Error
      Exit Code:    137
      Started:      Fri, 18 Oct 2024 18:23:55 +0000
      Finished:     Fri, 18 Oct 2024 18:27:41 +0000
    Ready:          True
    Restart Count:  1
    Liveness:       exec [cat /tmp/healthy] delay=5s timeout=30s period=5s #success=1 #failure=3
    Environment:    <none>
    Mounts:
      /var/run/secrets/kubernetes.io/serviceaccount from kube-api-access-ntqtc (ro)
.
.
.
Events:
  Type     Reason     Age                  From               Message
  ----     ------     ----                 ----               -------
  Normal   Scheduled  4m19s                default-scheduler  Successfully assigned default/example-15 to node01
  Normal   Pulled     4m14s                kubelet            Successfully pulled image "ubuntu" in 3.758s (3.758s including waiting). Image size: 29754422 bytes.
  Warning  Unhealthy  58s (x3 over 68s)    kubelet            Liveness probe failed: cat: /tmp/healthy: No such file or directory
  Normal   Killing    58s                  kubelet            Container container0 failed liveness probe, will be restarted
  Normal   Pulling    28s (x2 over 4m18s)  kubelet            Pulling image "ubuntu"
  Normal   Created    28s (x2 over 4m14s)  kubelet            Created container container0
  Normal   Started    28s (x2 over 4m14s)  kubelet            Started container container0
  Normal   Pulled     28s                  kubelet            Successfully pulled image "ubuntu" in 400ms (400ms including waiting). Image size: 29754422 bytes.