devops-exercises/topics/kubernetes/CKA.md

CKA (Certified Kubernetes Administrator)

• CKA (Certified Kubernetes Administrator)
  • Setup
  • Pods
    • Troubleshooting Pods
  • Namespaces
  • Nodes
  • Services
  • ReplicaSets
    • Troubleshooting ReplicaSets
  • Deployments
    • Troubleshooting Deployments
  • Scheduler
  • Labels and Selectors
  • Taints

Setup

• Set up Kubernetes cluster. Use on of the following

  1. Minikube for local free & simple cluster
  2. Managed Cluster (EKS, GKE, AKS)

• Set aliases

alias k=kubectl
alias kd=kubectl delete
alias kds=kubectl describe
alias ke=kubectl edit
alias kr=kubectl run
alias kg=kubectl get

Pods

Run a command to view all the pods in the current namespace

kubectl get pods

Note: create an alias (alias k=kubectl) and get used to k get po

Run a pod called "nginx-test" using the "nginx" image

k run nginx-test --image=nginx

Assuming you have a Pod called "nginx-test", how to remove it?

k delete nginx-test

In what namespace the etcd pod is running? list the pods in that namespace

k get po -n kube-system

Let's say you didn't know in what namespace it is. You could then run k get po -A | grep etc to find the Pod and see in what namespace it resides.

List pods from all namespaces

k get po -A

The long version would be kubectl get pods --all-namespaces.

Write a YAML of a Pod with two containers and use the YAML file to create the Pod (use whatever images you prefer)

cat > pod.yaml <<EOL
apiVersion: v1
kind: Pod
metadata:
  name: test
spec:
  containers:
  - image: alpine
    name: alpine
  - image: nginx-unprivileged
    name: nginx-unprivileged
EOL

k create -f pod.yaml

If you ask yourself how would I remember writing all of that? no worries, you can simply run kubectl run some_pod --image=redis -o yaml --dry-run=client > pod.yaml. If you ask yourself "how am I supposed to remember this long command" time to change attitude ;)

Create a YAML of a Pod without actually running the Pod with the kubectl command (use whatever image you prefer)

k run some-pod -o yaml --image nginx-unprivileged --dry-run=client > pod.yaml

How to test a manifest is valid?

with --dry-run flag which will not actually create it, but it will test it and you can find this way any syntax issues.

k create -f YAML_FILE --dry-run

How to check which image a certain Pod is using?

k describe po <POD_NAME> | grep -i image

How to check how many containers run in signle Pod?

k get po POD_NAME and see the number under "READY" column.

You can also run k describe po POD_NAME

Run a Pod called "remo" with the the latest redis image and the label 'year=2017'

k run remo --image=redis:latest -l year=2017

List pods and their labels

k get po --show-labels

Delete a Pod called "nm"

k delete po nm

List all the pods with the label "env=prod"

k get po -l env=prod

To count them: k get po -l env=prod --no-headers | wc -l

Troubleshooting Pods

You try to run a Pod but see the status "CrashLoopBackOff". What does it means? How to identify the issue?

The container failed to run (due to different reasons) and Kubernetes tries to run the Pod again after some delay (= BackOff time).

Some reasons for it to fail:

• Misconfiguration - mispelling, non supported value, etc.
• Resource not available - nodes are down, PV not mounted, etc.

Some ways to debug:

1. kubectl describe pod POD_NAME
   1. Focus on State (which should be Waiting, CrashLoopBackOff) and Last State which should tell what happened before (as in why it failed)
2. Run kubectl logs mypod
   1. This should provide an accurate output of
   2. For specific container, you can add -c CONTAINER_NAME
3. If you still have no idea why it failed, try kubectl get events

What the error ImagePullBackOff means?

Most likely you didn't write correctly the name of the image you try to pull and run. Or perhaps it doesn't exists in the registry.

You can confirm with kubectl describe po POD_NAME

How to check on which node a certain Pod is running?

k get po POD_NAME -o wide

Run the following command: kubectl run ohno --image=sheris. Did it work? why not? fix it without removing the Pod and using any image you want

Because there is no such image sheris. At least for now :)

To fix it, run kubectl edit ohno and modify the following line - image: sheris to - image: redis or any other image you prefer.

You try to run a Pod but it's in "Pending" state. What might be the reason?

One possible reason is that the scheduler which supposed to schedule Pods on nodes, is not running. To verify it, you can run kubectl get po -A | grep scheduler or check directly in kube-system namespace.

Namespaces

List all the namespaces

k get ns

Create a namespace called 'alle'

k create ns alle

Check how many namespaces are there

k get ns --no-headers | wc -l

Check how many pods exist in the "dev" namespace

k get po -n dev

Create a pod called "kartos" in the namespace dev. The pod should be using the "redis" image.

If the namespace doesn't exist already: k create ns dev

k run kratos --image=redis -n dev

You are looking for a Pod called "atreus". How to check in which namespace it runs?

k get po -A | grep atreus

Nodes

Run a command to view all nodes of the cluster

kubectl get nodes

Note: create an alias (alias k=kubectl) and get used to k get no

Create a list of all nodes in JSON format and store it in a file called "some_nodes.json"

k get nodes -o json > some_nodes.json

Services

Check how many services are running in the current namespace

k get svc

Create an internal service called "sevi" to expose the app 'web' on port 1991

How to reference by name a service called "app-service" within the same namespace?

app-service

How to check the TargetPort of a service?

k describe svc <SERVICE_NAME>

How to check what endpoints the svc has?

k describe svc <SERVICE_NAME>

How to reference by name a service called "app-service" within a different namespace, called "dev"?

app-service.dev.svc.cluster.local

Assume you have a deployment running and you need to create a Service for exposing the pods. This is what is required/known:

• Deployment name: jabulik
• Target port: 8080
• Service type: NodePort
• Selector: jabulik-app
• Port: 8080

kubectl expose deployment jabulik --name=jabulik-service --target-port=8080 --type=NodePort --port=8080 --dry-run=client -o yaml -> svc.yaml

vi svc.yaml (make sure selector is set to jabulik-app)

k apply -f svc.yaml

ReplicaSets

How to check how many replicasets defined in the current namespace?

k get rs

You have a replica set defined to run 3 Pods. You removed one of these 3 pods. What will happen next? how many Pods will there be?

There will still be 3 Pods running theoretically because the goal of the replica set is to ensure that. so if you delete one or more Pods, it will run additional Pods so there are always 3 Pods.

How to check which container image was used as part of replica set called "repli"?

k describe rs repli | grep -i image

How to check how many Pods are ready as part of a replica set called "repli"?

k describe rs repli | grep -i "Pods Status"

How to delete a replica set called "rori"?

k delete rs rori

How to modify a replica set called "rori" to use a different image?

k edis rs rori

Scale up a replica set called "rori" to run 5 Pods instead of 2

k scale rs rori --replicas=5

Scale down a replica set called "rori" to run 1 Pod instead of 5

k scale rs rori --replicas=1

Troubleshooting ReplicaSets

Fix the following ReplicaSet definition

apiVersion: apps/v1
kind: ReplicaCet
metadata:
  name: redis
  labels:
    app: redis
    tier: cache
spec:
  selector:
    matchLabels:
      tier: cache
  template:
    metadata:
      labels:
        tier: cachy
    spec:
      containers:
      - name: redis
        image: redis

kind should be ReplicaSet and not ReplicaCet :)

Fix the following ReplicaSet definition

apiVersion: apps/v1
kind: ReplicaSet
metadata:
  name: redis
  labels:
    app: redis
    tier: cache
spec:
  selector:
    matchLabels:
      tier: cache
  template:
    metadata:
      labels:
        tier: cachy
    spec:
      containers:
      - name: redis
        image: redis

The selector doesn't match the label (cache vs cachy). To solve it, fix cachy so it's cache instead.

Deployments

How to list all the deployments in the current namespace?

k get deploy

How to check which image a certain Deployment is using?

k describe deploy <DEPLOYMENT_NAME> | grep image

Create a file definition/manifest of a deployment called "dep", with 3 replicas that uses the image 'redis'

k create deploy dep -o yaml --image=redis --dry-run=client --replicas 3 > deployment.yaml

Remove the deployment `depdep`

k delete deploy depdep

Troubleshooting Deployments

Fix the following deployment manifest

apiVersion: apps/v1
kind: Deploy
metadata:
  creationTimestamp: null
  labels:
    app: dep
  name: dep
spec:
  replicas: 3
  selector:
    matchLabels:
      app: dep
  strategy: {}
  template:
    metadata:
      creationTimestamp: null
      labels:
        app: dep
    spec:
      containers:
      - image: redis
        name: redis
        resources: {}
status: {}

Change kind: Deploy to kind: Deployment

Fix the following deployment manifest

apiVersion: apps/v1
kind: Deployment
metadata:
  creationTimestamp: null
  labels:
    app: dep
  name: dep
spec:
  replicas: 3
  selector:
    matchLabels:
      app: depdep
  strategy: {}
  template:
    metadata:
      creationTimestamp: null
      labels:
        app: dep
    spec:
      containers:
      - image: redis
        name: redis
        resources: {}
status: {}

The selector doesn't match the label (dep vs depdep). To solve it, fix depdep so it's dep instead.

Scheduler

How to schedule a pod on a node called "node1"?

k run some-pod --image=redix -o yaml --dry-run=client > pod.yaml

vi pod.yaml and add:

spec:
  nodeName: node1

k apply -f pod.yaml

Note: if you don't have a node1 in your cluster the Pod will be stuck on "Pending" state.

Labels and Selectors

How to list all the Pods with the label "app=web"?

k get po -l app=web

How to list all objects labeled as "env=staging"?

k get all -l env=staging

How to list all deployments from "env=prod" and "type=web"?

k get deploy -l env=prod,type=web

Taints

Check if there are taints on node "master"

k describe no master | grep -i taints

Create a taint on one of the nodes in your cluster with key of "app" and value of "web" and effect of "NoSchedule"

k taint node minikube app=web:NoSchedule