K8s&helm&GKE@GCP

Create GKE cluster

0. Get GCP account

1. Open Cloud Shell

gcloud projects list
gcloud config set project {Project-Name}
gcloud config set compute/zone asia-east1-b
gcloud container clusters create cool-chicken

GKE release notes

其他配置說明可參考這裡

也可以參考官方文件，輸入 gcloud container clusters create --help 即可閱讀

這樣就可以成功創出一個 GKE cluster

Connect to GKE

Install gcloud

# Create environment variable for correct distribution
export CLOUD_SDK_REPO="cloud-sdk-$(lsb_release -c -s)"

# Add the Cloud SDK distribution URI as a package source
echo "deb http://packages.cloud.google.com/apt $CLOUD_SDK_REPO main" | sudo tee -a /etc/apt/sources.list.d/google-cloud-sdk.list

# Import the Google Cloud Platform public key
curl https://packages.cloud.google.com/apt/doc/apt-key.gpg | sudo apt-key add -

# Update the package list and install the Cloud SDK
sudo apt-get update && sudo apt-get install google-cloud-sdk

# Login and init gcloud
gcloud init

gcloud container clusters get-credentials cool-chicken \
--zone asia-east1-b \
--project indigo-night-266711

login to gcloud with {YOUR_ACCOUNT}

Once you successfully init gcloud, you can use gcloud container clusters get-credentials.... to login your k8s every time.

How to check your kubectl config

cat ~/.kube/config 

kubectl

k8s 控制台

Offical doc

sudo apt-get update && sudo apt-get install -y apt-transport-https
curl -s https://packages.cloud.google.com/apt/doc/apt-key.gpg | sudo apt-key add -
echo "deb https://apt.kubernetes.io/ kubernetes-xenial main" | sudo tee -a /etc/apt/sources.list.d/kubernetes.list
sudo apt-get update
sudo apt-get install -y kubectl

Set alias for kubectl

alias k='kubectl'

Create your first pod

Create your namespace

kubectl create namespace ${MY_NAMESPACE_NAME}
kubectl get namespace

Switch to your namespace

kubectl config set-context --current --namespace ${MY_NAMESPACE_NAME}

Query the resource under your namespace

# 查看預設namespace下的所有pod
kubectl get pod

# 查看kube-system namespace下的所有pod
kubectl get pod -n kube-system

# 查看所有namespace下的所有pod
kubectl get pod -A

Create a random-logger in your namespace

kubectl run random-logger --image=chentex/random-logger

Check the logger in your random-logger

kubectl get deployment
kubectl get pod
kubectl logs -f {POD_NAME}

Delete pod

kubectl delete pod {POD_NAME}
kubectl get pod

Why pod still exist? Because deployment restart it.
Let's delete the deployment

kubectl get deployment
kubectl delete deployment {DEPLOYMENT_NAME}

If you want to interact with Pod,
try kubectl run ubuntu -it --rm --image=ubuntu:16.04

Debugging with kubectl

kubectl exec -i -t {POD_NAME} -- sh

kubectl attach {POD_NAME}

exec 會執行 pod 內的 container 的指定程式
attach 會進入 pod 內 container 的 main process

Create a http server

kubectl run http --image=katacoda/docker-http-server:latest --port=80

kubectl port-forward {POD_NAME} 8000:80

curl localhost:8000

Create a http server with service

kubectl expose 會為 deployment 建立一個 service

kubectl expose deployment http --target-port=80 --type=NodePort
kubectl get service 

Check service load balance

kubectl scale deployment http --replicas=3
kubectl get deployment http
kubectl describe service http # Endpoints: (contains 3 endpoints)

Service 後面可以對應多個 deployment，但通常都一對一。

test load balance with busybox

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

ping http # cluster-ip 172.20.102.166 will be shown
wget -O - http
wget -O - http
wget -O - http

EKS 上的芳鄰

每個 deployment 在該 EKS cluster 的內部都會自動註冊 DNS，所有該 cluster 內的 pod 都可以以此連線到其他 pod。

e.g. cross namespace access <service-name>.<namespace-name>.svc.cluster.local
In busybox:

ping http.{YOUR_NAMESPACE}.svc.cluster.local
wget -O - http.{YOUR_NAMESPACE}.svc.cluster.local
wget -O - http.{YOUR_NAMESPACE}.svc.cluster.local
wget -O - http.{YOUR_NAMESPACE}.svc.cluster.local
exit

這部分不用設 service 也會生效

deploy resource with yaml

使用yaml manifest部屬

---

Deployment manifest

# deployment.yaml
apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: webapp1 # deployment name
spec:
  replicas: 1
  template:
    metadata:
      labels:
        app: webapp1 # service selector
    spec:
      containers:
      - name: webapp1 # pod name
        image: katacoda/docker-http-server:latest
        ports:
        - containerPort: 80

kubectl create -f deployment.yaml
kubectl get deployment
kubectl describe deployment webapp1

---

Service manifest

# service.yaml
apiVersion: v1
kind: Service
metadata:
  name: webapp1
spec:
  type: NodePort
  ports:
  - port: 80
  selector:
    app: webapp1  # service selector

kubectl create -f service.yaml
kubectl get service
kubectl describe service webapp1

---

Ingress manifest

# ingress.yaml
apiVersion: extensions/v1beta1
kind: Ingress
metadata:
  name: webapp1
spec:
  backend:
    serviceName: webapp1
    servicePort: 80

kubectl create -f ingress.yaml
kubectl get ingress

You can check the status of http-balancer in here.

Test your webapp

Open browser

http://{ADDRESS}

How ingress work

NGINS is for aws, google use their own http-balancer

Helm

install helm

# install with binary
wget https://get.helm.sh/helm-v3.0.2-linux-amd64.tar.gz
mkdir ~/download/helm
tar -C ~/download/helm -xzf helm-v3.0.2-linux-amd64.tar.gz
mv ~/download/helm/linux-amd64/helm /usr/local/bin/
# Add repo list
helm repo add stable https://kubernetes-charts.storage.googleapis.com/

install with local helm chart

git clone https://github.com/helm/charts.git
cd charts/stable/hackmd
helm dependency update .
helm install hackmd .

helm install stable/hackmd hackmd
kubectl port-forward {YOUR_POD} 8080:3000

reference

• k8s Offical doc
• Google GKE doc
• k8s workshop Pt 0
• Day 27 - 漫步在雲端：在GCP 中建立k8s 叢集 - iT 邦幫忙

kube ops view