e.observability.md

Observability (18%)

Liveness and readiness probes

kubernetes.io > Documentation > Tasks > Configure Pods and Containers > Configure Liveness and Readiness Probes

Create an nginx pod with a liveness probe that just runs the command 'ls'. Save its YAML in pod.yaml. Run it, check its probe status, delete it.

show

kubectl run nginx --image=nginx --restart=Never --dry-run -o yaml > pod.yaml
vi pod.yaml

apiVersion: v1
kind: Pod
metadata:
  creationTimestamp: null
  labels:
    run: nginx
  name: nginx
spec:
  containers:
  - image: nginx
    imagePullPolicy: IfNotPresent
    name: nginx
    resources: {}
    livenessProbe: # our probe
      exec: # add this line
        command: # command definition
        - ls # ls command
  dnsPolicy: ClusterFirst
  restartPolicy: Never
status: {}

kubectl create -f pod.yaml
kubectl describe pod nginx | grep -i liveness # run this to see that liveness probe works
kubectl delete -f pod.yaml

Modify the pod.yaml file so that liveness probe starts kicking in after 5 seconds whereas the interval between probes would be 5 seconds. Run it, check the probe, delete it.

show

kubectl explain pod.spec.containers.livenessProbe # get the exact names

apiVersion: v1
kind: Pod
metadata:
  creationTimestamp: null
  labels:
    run: nginx
  name: nginx
spec:
  containers:
  - image: nginx
    imagePullPolicy: IfNotPresent
    name: nginx
    resources: {}
    livenessProbe: 
      initialDelaySeconds: 5 # add this line
      periodSeconds: 5 # add this line as well
      exec:
        command:
        - ls
  dnsPolicy: ClusterFirst
  restartPolicy: Never
status: {}

kubectl create -f pod.yaml
kubectl describe po nginx | grep -i liveness
kubectl delete -f pod.yaml

Create an nginx pod (that includes port 80) with an HTTP readinessProbe on path '/' on port 80. Again, run it, check the readinessProbe, delete it.

show

kubectl run nginx --image=nginx --dry-run -o yaml --restart=Never --port=80 > pod.yaml
vi pod.yaml

apiVersion: v1
kind: Pod
metadata:
  creationTimestamp: null
  labels:
    run: nginx
  name: nginx
spec:
  containers:
  - image: nginx
    imagePullPolicy: IfNotPresent
    name: nginx
    resources: {}
    ports:
      - containerPort: 80 # Note: Readiness probes runs on the container during its whole lifecycle. Since nginx exposes 80, containerPort: 80 is not required for readiness to work.
    readinessProbe: # declare the readiness probe
      httpGet: # add this line
        path: / #
        port: 80 #
  dnsPolicy: ClusterFirst
  restartPolicy: Never
status: {}

kubectl create -f pod.yaml
kubectl describe pod nginx | grep -i readiness # to see the pod readiness details
kubectl delete -f pod.yaml

Logging

Create a busybox pod that runs 'i=0; while true; do echo "$i: $(date)"; i=$((i+1)); sleep 1; done'. Check its logs

show

kubectl run busybox --image=busybox --restart=Never -- /bin/sh -c 'i=0; while true; do echo "$i: $(date)"; i=$((i+1)); sleep 1; done'
kubectl logs busybox -f # follow the logs

Debugging

Create a busybox pod that runs 'ls /notexist'. Determine if there's an error (of course there is), see it. In the end, delete the pod

show

kubectl run busybox --restart=Never --image=busybox -- /bin/sh -c 'ls /notexist'
# show that there's an error
kubectl logs busybox
kubectl describe po busybox
kubectl delete po busybox

Create a busybox pod that runs 'notexist'. Determine if there's an error (of course there is), see it. In the end, delete the pod forcefully with a 0 grace period

show

kubectl run busybox --restart=Never --image=busybox -- notexist
kubectl logs busybox # will bring nothing! container never started
kubectl describe po busybox # in the events section, you'll see the error
# also...
kubectl get events | grep -i error # you'll see the error here as well
kubectl delete po busybox --force --grace-period=0

Get CPU/memory utilization for nodes (metrics-server must be running)

show

kubectl top nodes