Skip to content

Latest commit

 

History

History
359 lines (242 loc) · 9.34 KB

02-pods-deployments-and-services.md

File metadata and controls

359 lines (242 loc) · 9.34 KB
Raw

Lab 02 - Pods, Deployments and Services

Prerequisites

Initial concepts

Pods

A Pod represents a set of running containers on your cluster - it's the smallest and simplest Kubernetes object.

Creating and updating Pods

  1. Using kubectl, create a new Pod called 'webserver' running nginx image.

    kubectl run webserver --image=nginx --restart=Never

  2. Get the status and IP of Pod 'webserver'.

    kubectl get pod webserver -o wide

  3. Get a detailed description of the pod, including related resources such as events or controllers.

    kubectl describe pod webserver

  4. Get the status of 'faultywebserver'.

    kubectl get pod faultywebserver -o wide

    Notice that the Pod status is "ImagePullBackOff". Something is wrong.

  5. Identify, and resolve, the issue with 'faultywebserver'.

    kubectl describe pod faultywebserver 
kubectl set image pod faultywebserver faultywebserver=nginx
kubectl get pod faultywebserver -o wide

  6. Using YAML, create a new Pod called 'web' running nginx image.

    # use kubectl to generate web-pod.yaml
kubectl run webserver --image=nginx --restart=Never --dry-run -o yaml > web-pod.yaml

# view nginx-pod.yaml
cat web-pod.yaml

# apply the YAML file
kubectl apply -f web-pod.yaml

    Notice the use of --dry-run -o yaml to generate the yaml, which is then piped into web-pod.yaml.

Labels and annotations

  1. Using kubectl, create a new Pod called 'nginx' running nginx image. Label the Pod with 'app=v1'.

    kubectl run nginx --image=nginx --restart=Never --labels=app=v1

# or

kubectl run nginx --image=nginx --restart=Never
kubectl label pod nginx app=v1

  2. Show all Pods in the default namespace, including the Pods' labels.

    kubectl get pod --show-labels

  3. Change the label of pod 'nginx' to be 'app=v2'

    kubectl label pod nginx app=v2 --overwrite

  4. Get all Pods with the label 'app=v2'.

    kubectl get pod -l app=v2

  5. Remove the 'app' label from Pod 'nginx'.

    kubectl label pod nginx app-

  6. Annotate Pod 'nginx' with "description='my description'".

    kubectl annotate pod nginx description='my description'

Logs and Exec

  1. Get logs from 'webserver'.

    kubectl logs webserver

  2. Retrieve a list of all the files within the nginx container running on 'webserver' pod.

    kubectl exec webserver -- ls

Deployments

A deployment is an API object that manages a replicated application. A Deployment controller provides declarative updates for Pods and ReplicaSets.

  1. Create a yaml file for a Deployment running a nginx container, with five replicas, and save it as mydeploy.yaml.

    kubectl run mydeploy --image=nginx --replicas=5 --dry-run -o yaml > mydeploy.yaml

    This is another great example where you can use --dry-run -o yaml to generate the required yaml.

  2. Create a Deployment using mydeploy.yaml.

    kubectl apply -f mydeploy.yaml

  3. View the Deployment 'mydeploy', the associated ReplicaSet and Pods.

    kubectl get deployment,rs,pod

  4. Modify the image used by 'mydeploy' to use image nginx:1.16.0 and observe the update as it's applied.

    # set the deployment image
kubectl set image deployment mydeploy mydeploy=nginx:1.16.0

# observe how the update gets applied 
kubectl get rs -w

  5. View the rollout history for 'mydeploy' and roll back to the previous revision.

    # view previous rollout revisions and configurations.
kubectl rollout history deploy mydeploy

# rollback to the previous rollout.
kubectl rollout undo deploy mydeploy

# observe how rollowing update gets applied 
kubectl get rs -w

  6. Scale 'mydeploy' to 2 instance.

    kubectl scale deploy mydeploy --replicas=2

Services

A Service is an abstract way to expose an application running on a set of Pods.

  1. Expose the deployment 'mydeploy' on port 80. Observe that a service is created.

    kubectl expose deployment mydeploy --port 80
kubectl get svc mydeploy

  2. Confirm that a Cluster IP has been created.

    kubectl get svc mydeploy -o jsonpath='{.spec.clusterIP}'

  3. Using the Pod's Cluster IP, create a new temporary Pod using busybox and 'hit' the IP with wget:

    # get the service's Cluster IP
kubectl get svc mydeploy -o jsonpath='{.spec.clusterIP}'

# run busybox
kubectl run busybox --rm --image=busybox -it --restart=Never -- sh

# from inside the container
wget -O- <cluster-ip>:80
exit

  4. Change to service type of 'mydeploy' from the ClusterIP to LoadBalancer. Find the Public IP address and browse to the application.

    # edit the service
kubectl edit svc mydeploy

# change "type: ClusterIP" to "type: LoadBalancer"

# get the assigned external IP
kubectl get svc -w

    Note: This may take a few minutes to complete.

Fireworks scenario

SignalR based application that allows website users to light fireworks and display on all the connected site users. You can light single or multi shot using the app. There is also a button that can stimulate a crash /home/admin. Pressing the button again will make the application run again.

Port Exposed:

  • 80

Images on Docker Hub:

  • Green: kunalbabre/fireworks:green
  • Blue: kunalbabre/fireworks:blue
  • Red: kunalbabre/fireworks:red
  • Yellow: kunalbabre/fireworks:yellow

Trigger fireworks manually:

  • Trigger Single: /home/singleshot
  • Trigger Multishot: /home/multishot

Environment variables:

  • SIGNALR_CS: (optional) if you wish to scale-out you can provide connection string for Redis or Azure SignalR
  • APP_COLOR: (works with latest tag) you can specify theme color for the app (red,green, blue, yellow)

Health Monitoring:

  • Liveness: /home/isRunning

    • returns HTTP 200 if the application is alive
    • returns HTTP 400 if the application has crashed

  • Readiness: /home/isRunning

    • returns HTTP 200 if the application is alive
    • returns HTTP 400 if the application has crashed

Core Goal

  1. All operations in this exercise should be performed in the fireworks namespace (already been created for you).

    hint 

    kubectl config set-context $(kubectl config current-context) --namespace=<namespace>

  2. Create Deployment called 'fireworks' using the image kunalbabre/fireworks:red in the namespace 'fireworks'.

    hint 

    kubectl create deployment <name> --image <image name>

  3. Expose the deployment firework on port 80. Use a LoadBalancer service called 'fireservice'.

    hint 

    kubectl expose deployment <deployment-name> --name <service-name> --port=80 --type <service-type>

  4. Wait for the external IP to be allocated. Once available, navigate to the IP address in the web browser.

    hint 

    kubectl get svc -w

  5. Scale the deployment 'fireworks' to 5 replicas. Observe the Pods being created.

    hint 

    # using edit command to update replica count
kubectl edit deployment <deployment name>

# or, using the scale command
kubectl scale deployment <deployment name> --replicas=5

# finally, you can watch pods being created
kubectl get po -w

  6. Update the deployment 'fireworks' to use image kunalbabre/fireworks:latest.

    hint 

    # using edit command to update container image
kubectl edit deployment <deployment name>

# or, using set image command
kubectl set image deploy <deployment name> fireworks=<New Image Name>

Stretch Goal

  1. Configure the 'fireworks' pods to only accept traffic when ready and auto restart if crashed.

    hint

    Look for examples in the Kubernetes documentation.