In this task you will learn how to describe and inspect a simple application as a kubernetes stateful set and expose it to the outside world.

• Application details
• Task overview
• Useful commands

1. Labels and annotations
2. Environment variables
3. Pod health checks
4. Pod resource limits
5. Service
6. Ingress

We have prepared a workshop application container with the following features:

1. The containerized application is an http server listening on HTTP port 80.
2. There are several http paths exposed within the container:
   • / - serves Space Invaders
   • /hello - this path will respond with Hello, $CODE message, where $CODE is an environment variable set in the definition of a pod.
   • /live - Health check liveness probe , where application reports its current status. While the application is running it returns HTTP 200 code. If the application has been running for more than ten minutes, it will respond with HTTP 500 error.
   • /ready - Health check readiness probe, where application reports if it is ready to accept web request. If the application is ready to accept traffic, this path will respond with HTTP 200 response.
3. Application resource requirements:
   • Between 10m and 100m CPU.
   • Between 32M and 128M Memory.

Current directory contains three kubernetes objects that are stripped to bare working minimum.

Objects:

1. StatefulSet - Object containing application definition. This is the file we will be editing in workshop tasks.
2. Service - Object used to expose the application pods internally within the cluster.
3. Ingress - Object to expose the application externally on the URL https://$CODE.k8s.3fs.si where $CODE is you personalized workshop code.

Your task is to extend the objects with additional options to give Kubernetes more information how to run your application and to align with best practices.

See kubectl cheat sheet for full command reference.

kubectl get all
kubectl describe statefulset/workshop-example
kubectl describe service/workshop-example
kubectl describe ingress/workshop-example
kubectl describe pod/workshop-example-0
kubectl logs -f pod/workshop-example-0
kubectl apply -f service.yaml
kubectl replace --force -f statefulset.yaml

Current Pod definition in statefulset.yaml lacks definition of annotations.

1. Add the following annotations to Pod definition (yaml: .spec.template.metadata.annotations):

   • workshop-name: k8s-workshop-episode-2
   • any annotation of your choice

2. Update Pod definition with the following labels (yaml: .spec.template.metadata.labels):

   • code: $CODE, where $CODE is your workshop code

Hint: use example in statefulset.yaml

After you have updated the statefulset.yaml, deploy the application with the following command.

kubectl apply -f statefulset.yaml

Inspect the deployed objects by using the following kubectl commands:

kubectl get all
kubectl describe statefulset/workshop-example
kubectl describe pod/workshop-example-0
kubectl logs pod/workshop-example-0

Select deployed objects filtering using your label.

kubectl get all -l code=$CODE

--- statefulset.yaml	2020-05-26 17:38:05.000000000 +0200
+++ solutions/01_statefulset.yaml	2020-05-26 17:25:32.000000000 +0200
@@ -14,31 +14,18 @@ spec:
       name: workshop-app-example
   template:
     metadata:
-      annotations: {}
+      annotations:
+        workshop: k8s-workshop-episode-2
+        date: "2020-05-27"
       labels:
         name: workshop-app-example
+        code: cranky-hippo
     spec:
       containers:
         - name: workshop-app-example

kubectl replace --force -f statefulset.yaml

The application is using environment variables for configuration. The environment variable CODE is displayed on HTTP endpoint /hello. You can check that response by visiting https://$CODE.k8s.3fs.si/hello.

Add the environment variables:

• CODE with the value of your workshop code to Pod definition of the StatefulSet (yaml: .spec.template.spec.containers[0].env)
• WORKSHOP_USERNAME with custom value k8s-workshop. This value will be used in the future tasks.

Hint: use example in statefulset.yaml

After you have updated the statefulset.yaml, deploy the application with the following command.

kubectl replace --force -f statefulset.yaml

Inspect the deployed object by using kubectl describe commands.

kubectl describe statefulset/workshop-example
kubectl describe pod/workshop-example-0

Check pods environment by entering the running pod using kubectl exec command and running env command.

kubectl exec pod/workshop-example-0 env

--- 01_statefulset.yaml	2020-05-26 17:59:28.000000000 +0200
+++ 02_statefulset.yaml	2020-05-26 17:25:32.000000000 +0200
@@ -25,8 +25,12 @@ spec:
         - name: workshop-app-example
           image: "eu.gcr.io/k8s-workshop-2/app:latest"
           imagePullPolicy: IfNotPresent
-          env: []
           ports:
             - name: http
               containerPort: 80
               protocol: TCP
+          env:
+            - name: CODE
+              value: "cranky-hippo"
+            - name: WORKSHOP_USERNAME
+              value: "k8s-workshop"

kubectl replace --force -f statefulset.yaml

The application has no health check configured.

Add health check definitions to Pod definition in StatefulSet:

• add livenessProbe with the following definition (yaml key: .spec.template.spec.containers[0].livenessProbe)

  livenessProbe:
    httpGet:
      path: /live
      port: http

• add readinessProbe with the following definition (yaml key: .spec.template.spec.containers[0].readinessProbe)

  readinessProbe:
      httpGet:
        path: /ready
        port: http

Hint: use example in statefulset.yaml

After you have updated the statefulset.yaml, deploy the application with the following command.

kubectl replace --force -f statefulset.yaml

Inspect the pod/workshop-example-0 object by using kubectl describe pod/workshop-example-0 command.

kubectl describe statefulset/workshop-example
kubectl describe pod/workshop-example-0

--- 02_statefulset.yaml	2020-05-26 17:25:32.000000000 +0200
+++ 03_statefulset.yaml	2020-05-26 17:25:32.000000000 +0200
@@ -34,3 +34,11 @@ spec:
               value: "cranky-hippo"
             - name: WORKSHOP_USERNAME
               value: "cranky-hippo"
+          livenessProbe:
+            httpGet:
+              path: /live
+              port: http
+          readinessProbe:
+            httpGet:
+              path: /ready
+              port: http

kubectl replace --force -f statefulset.yaml

Current definition of StatefulSet lacks information about resource usage.

Update definition of Pod in StatefulSet with the following limits:

• when the container is starting, it should requests 10m of CPU time and 32M of memory.
• the container should be limited to 100m of CPU time and 128M of memory

Use the following definitions in .spec.template.spec.containers[0].resources:

  requests:
    cpu: 10m
    memory: 32Mi
  limits:
    cpu: 100m
    memory: 128Mi

Within the workshop console container updated StatefulSet definition to the cluster:

kubectl replace --force -f statefulset.yaml

Inspect the pod/workshop-example-0 object with kubectl command.

kubectl describe pod/workshop-example-0

--- 03_statefulset.yaml	2020-05-26 17:25:32.000000000 +0200
+++ 04_statefulset.yaml	2020-05-26 17:25:43.000000000 +0200
@@ -42,3 +42,10 @@ spec:
             httpGet:
               path: /ready
               port: http
+          resources:
+            requests:
+              cpu: 10m
+              memory: 32Mi
+            limits:
+              cpu: 100m
+              memory: 128Mi

kubectl replace --force -f statefulset.yaml

Expose the application inside Kubernetes cluster by deploying a service. It is defined in service.yaml.

Pods running in Kubernetes will now be able to talk to your application on port 80 using the service.

kubectl describe service/workshop-example

kubectl apply -f service.yaml

To expose the service to Internet Ingress object needs to be created. Additionally, if we want to use HTTPS, the certificate has to be present in the kubernetes environment. The certificate must be stored in kubernetes Secret object and referenced in Ingress definition.

A .yaml file to create the wildcard certificate secret is located in the / folder of Docker container. This object will be important in the future tasks. Apply it using the following command:

kubectl create -f /k8s.3fs.si-cert.yaml

After successful import of certificate Secret object, we can create Ingress object to expose service to Internet.

To expose the service to the Internet edit hostname in ingress.yaml to include $CODE (e.g. cranky-hippo.k8s.3fs.si). To edit file you can use included vim or nano editors (quick guide how to use them is here).

The spec section should look similar to:

  tls:
    - hosts:
        - "cranky-hippo.k8s.3fs.si"
      secretName: k8s.3fs.si-certificate
  rules:
    - host: "cranky-hippo.k8s.3fs.si"
      http:
        paths:
          - path: /
            backend:
              serviceName: workshop-example
              servicePort: 80

Create the Ingress object named workshop-example using the following command.

kubectl apply -f ingress.yaml

Inspect the ingress object.

kubectl describe ingress workshop-example

Your workshop application has now been deployed. Replace $CODE with your workshop code.

Access:

• Space invaders: https://$CODE.k8s.3fs.si.
• Hello: https://$CODE.k8s.3fs.si/hello.

Kubernetes dashboard:

• All deployed resources: https://dashboard.k8s.3fs.si/#/overview?namespace=$CODE
• Pod information: https://dashboard.k8s.3fs.si/#/pod/$CODE/workshop-example-0?namespace=$CODE
• Logs: https://dashboard.k8s.3fs.si/#/log/$CODE/workshop-example-0/pod?namespace=$CODE

Things to check:

• Pod CPU and Memory Resource consumption
• Pod Metadata
• Pod container environment variables
• Pod Events (hint: pod liveness probe will fail after 10min.)
• Pod Logs
• Service definition
• Ingress definition
• Secret used for HTTPS