Monitor¶

azure monitor¶

Best to enable azure monitor.
It provides valuable insights into the health and performance of your aks resources.
The specific cost will depend on the amount of data you collect and the retention period you choose.

Commands for monitoring¶

kubectl get <resource type> <resource name>
kubectl describe <resource type> <resource name>
kubectl logs <pod name>

kubectl get¶

Lists resources such as pods, ReplicaSets, ingresses, nodes, deployments, secrets, and so on.

#show all deployments, ReplicaSets, pods, and services
kubectl get all
#get pods status
kubectl get pods
#output with extra columns
kubectl get pods -o wide

kubectl describe¶

It contains the details of the object itself, as well as any recent events related to that object.

#just pods
kubectl describe pods
#a particular pod
kubectl describe pod/<pod-name> #or use space: kubectl describe pod <pod-name>
#get all events in a cluster
kubectl get events

debug app¶

image pull error¶

#create error
kubectl edit deployment/frontend
kubectl get pods
#get full error details
kubectl describe pods/<failed pod name>

app error¶

kubectl get service
#scale down frontend
kubectl scale --replicas=1 deployment/frontend
#launch bash shell on pod
kubectl exec -it <frontend-pod-name> -- bash
#install vim
apt update
apt install -y vim
#update guestbook.php
vim guestbook.php
#get logs
kubectl logs <frontend-pod-name>
#get live log stream
kubectl logs <pod-name> -f
#solve error
kubectl delete pod <podname>

probe¶

A liveness probe monitors the availability of an application while it is running. If a liveness probe fails, Kubernetes will restart your pod.

A readiness probe monitors when your application becomes available. If a readiness probe fails, Kubernetes will not send any traffic to the unready pods.

218