khool-sleepingbeauty

This is a sample project to look show the readiness process

requisit

This project assume that you have docker, kubernetes, a registry this can be resume with minikube

minikube -p sleepingbeauty start --cpus='4' --memory='16g' --disk-size=60g
eval $(minikube -p sleepingbeauty docker-env)
# create sleepingbeauty namespace and use it
kubectl create ns sleepingbeauty
kubectl config set-context --current --namespace=sleepingbeauty

build and deploy the project

since we use default goals

mvn

it will package k8s:resource k8s:apply, without test . therefore it should deploy 3 pods and a service.

The pod have a livenessProbe on /live and readinessProbe on /ready. It has two other endpoint :

/goodnight that will toogle ReadinessResource.sleeping to TRUE and therefore break the /ready endpoint
/wakeup that will toogle ReadinessResource.sleeping to FALSE and therefore restore the /ready endpoint

You can watch the pod via watch kubectl get pods or kubectl get pods -w

let play

look into the service

kubectl port-forward svc/sleepingbeauty 8080:80

you can put your pod to sleep here

kubectl get pods
    
# NAME                              READY   STATUS    RESTARTS   AGE
# sleepingbeauty-f6d6dddb7-4jk8k    0/1     Running   0          1m12s
# sleepingbeauty-f6d6dddb7-q5t6b    1/1     Running   0          1m1s
# sleepingbeauty-f6d6dddb7-z49z2    1/1     Running   0          1m6s

with you pod forward you are still linked to your pod (sleepingbeauty-f6d6dddb7-4jk8k), but you can see this it is not ready, so neworks are not push to it any more, If you restart your port-forward you'll be put to another pod but you can see that you do not have an extra pod, you have 3 pods one of which is not ready.
I order to help diagnose you can change you pod label so you will keep it for diagnostic while spawning a new one for your running project.

kubectl label po $(kgpo -l provider=jkube | grep '0/1' | cut -d ' ' -f 1) app- group- sick=yes

then check you have your 3 active pods

kubectl get pods
    
# NAME                              READY   STATUS    RESTARTS   AGE
# sleepingbeauty-f6d6dddb7-4jk8k    0/1     Running   0          2m12s
# sleepingbeauty-f6d6dddb7-j8nhs    1/1     Running   0          15s
# sleepingbeauty-f6d6dddb7-q5t6b    1/1     Running   0          2m1s
# sleepingbeauty-f6d6dddb7-z49z2    1/1     Running   0          2m6s

dive into you sick pod

Get yaml info :

kubectl get pods -l sick=yes -o name -o yaml

# apiVersion: v1
# items:
# - apiVersion: v1
#  kind: Pod
#  metadata:
#    annotations:
#      io.github.kanedafromparis/fun: "true"
#      jkube.io/git-branch: main
#      jkube.io/git-commit: 56943d3d0d87e764fc4c295f40e30cc18d646084
#      jkube.io/git-url: git@github.com:kanedafromparis/khool-sleepingbeauty.git
#    creationTimestamp: "2021-09-08T16:56:30Z"
#    generateName: sleepingbeauty-f6d6dddb7-
#    labels:
#      pod-template-hash: f6d6dddb7
#  ....

Check you pods log

kubectl logs -l sick=yes --tail=-1

Go into your pod (this practice is controversial, on the security point-of-view)

kubectl exec -it $(kubectl get pods -l sick=yes -o name) -- sh

kanedafromparis / khool-sleepingbeauty