Kubernetes Cluster

Kubernetes cluster

Creating clusters

#To setup the project and cluster

gcloud config set project peaceful-signer-194319

gcloud config set compute/zone us-central1-a

#get the Kubernetes Engine server config

gcloud container get-server-config --zone us-central1-a

#get only the default Kubernetes Engine server version

gcloud container get-server-config --zone us-central1-a pipe grep defaultClusterVersion:

#create (start) a cluster on Google Cloud Platform with 2 nodes and cluster version 1.9.2

gcloud container clusters create mycluster --num-nodes=2 --cluster-version 1.9.2-gke.1

#to get the cluster credentials and configure the kubectl command-line tool

gcloud container clusters get-credentials mycluster

#to start Minikune on windows

minikube start --kubernetes-version v1.8.0 --VM-driver hyperv --hyperv-virtual-switch "Primary Virtual Switch"

Resizing a cluster

#setup the project and cluster

gcloud config set project [PROJECT_ID]

gcloud config set compute/zone [COMPUTE/ZONE]

gcloud config set project peaceful-signer-194319

gcloud config set compute/zone us-central1-a

#to create(start) a cluster on GCP with 2 nodes using the default cluster version

gcloud container clusters create mycluster --num-nodes=2

#List existing clusters in the default zone for running containers

gcloud container clusters list

#Resize the cluster, We will change it to one single Node in the Node pool

gcloud container clusters resize mycluster --size 1

#To list the GCP K8 cluster

gcloud container clusters list

Stopping a Cluster

#Although you can’t really ‘stop’ a cluster on GCP you can ensure that no Pods are going

# to be deployed by setting the number of nodes to 0

gcloud container clusters resize mycluster –size=0

#To list the GCP K8 cluster

gcloud container clusters list

#for Minikube on windows

kubectl version

kubectl get all

minikube stop

Deleting a Cluster

# Delete an existing cluster

# --zone overrides the default COMPUTE/ZONE setting

# --async doesn’t wait for the operation to complete returns to the prompt immediately

# there are also GCLOUD WIDE FLAGS like –project, --quiet, --account and so on

glcoud container clusters delete mycluster –zone=us-centrall-a –async

#for Minikube

minikube stop

minikube delete

#rm -rf ~/.minikube for Linux

Upgrading a Cluster

# list existing clusters in the default zone

gcloud container clusters list

# to fine the supported Kubernetes master and node versions for upgrades

gcloud container get-server-config

#To upgrade the cluster version

gcloud container clusters upgrade mycluster --master --cluster-version 1.9.2-gke.1

#To upgrade nodes

gcloud container clusters upgrade mycluster

#minikube upgrade version

minikube get-k8-versions

minikube start --kubernetes-version v1.8.0 --VM-driver hyperv --hyperv-virtual-switch "Primary Virtual Switch"

minikube stop

minikube start --kubernetes-version v1.9.0 --VM-driver hyperv --hyperv-virtual-switch "Primary Virtual Switch"

kubectl version | sls "Server Version:"

Understanding Kubeconfig

#To setup the project and cluster

gcloud config set project peaceful-signer-194319

gcloud config set compute/zone us-central1-a

#get the Kubernetes Engine server config

gcloud container get-server-config --zone us-central1-a

#create (start) a cluster on Google Cloud Platform with 2 nodes and cluster version 1.9.2

gcloud container clusters create mycluster --num-nodes=2

gcloud container clusters create mycluster-dev --num-nodes=2

# here is where the default is kubeconfig file is located and it’s names config

ls ~/.kube

# if you want to look at the file, you can ‘cat ~/.kube/config’, just be aware that your

# certificate-authority-data will be in full view if you do

kubectl config view

#kubectl config SUBCOMMAND

kubectl config –kubeconfig=d-config get-clusters

kubectl config –kubeconfig=p-config get-clusters

kubectl config get-clusters

#kubernetes.io/docs/reference/generated/kubectl/kubectl-commands#config

Sharing a Cluster

kubectl get namespaces

# Start minikube

minikube start --kubernetes-version v1.9.0 --VM-driver hyperv --hyperv-virtual-switch "Primary Virtual Switch"

#alter your config located at ~/.kube/config

#create two new namespaces

kubectl create -f C:\os_kube\course4\121653\ns-dev.yaml

kubectl create -f C:\os_kube\course4\121653\ns-test.yaml

kubectl get namespaces –show-labels

#set some contexts in the configuration for the cluster

kubectl config set-context dev --namespace=dev --cluster=minikube –user=Joe-dev

kubectl config set-context test --namespace=test --cluster=minikube –user=Joe-test

kubectl config use-context dev

kubectl config current-context

#create a deployment in the dev context

kubectl run nginx --image=nginx –replicas=3

#now switch contexts to test, and see whether we can see the deployment

kubectl config use-context test

kubectl get all

ns-dev.yaml

------------

apiVersion:v1

kind: Namespace

metadata:

name: dev

labels:

name: dev

ns-test.yaml

------------

apiVersion:v1

kind: Namespace

metadata:

name: test

labels:

name: test

Authentication Clusters

#set project and cluster

gcloud config set project peaceful-signer-194319

gcloud config set compute/zone us-central1-a

# On GCP add a user with IAM and give user the role Kubernetes Engine Viewer

# let’s look at the iam policy for this project

gcloud projects get-iam-policy peaceful-signer-194319 --format json > iam.json

Creating a Alpha Cluster

# Alpha cluster are not for productions

# they can’t upgraded and automatically deleted after 30 days

# To create alpha cluster

gcloud alpha container clusters create alphacluster --enable-kubernetes-alpha –cluster-version=1.9.2-gke.1

# you can get the authentication credentials for the cluster

gcloud alpha container clusters get-credentials alphacluster

#later to list the cluster and see when the alpha clusters expire

gcloud alpha container clusters list

Using Cluster Autoscaler

#set project and cluster

gcloud config set project peaceful-signer-194319

gcloud config set compute/zone us-central1-a

# To create an autoscaling cluster use ‘cloud container clusters create’

# ‘--enable autoscaling’ enables autoscalin

# --min-nodes and --max-nodes are the minimum number of nodes and the maximum number of nodes

gcloud container clusters create mycluster --num-nodes 1 --enable-autoscaling --min-nodes 1 --max-nodes 5

#now let’s deploy to the deployment, removing all but 1 pod

# then check the deplyment and nodes

# you’ll see the nodes scale back in a few minues

kubectl create -f deployment.yaml

kubectl get nodes

kubectl get deploy

#now let’s apply a change to the deployment, removing all but 1 pod

# then check the deplyment and nodes

# you’ll see the nodes scale back in a few minues

kubectl apply -f alterdeployment.yaml

kubectl get nodes

kubectl get deploy

# To disable autoscaling for a specific node pool

# ‘--no--enable-autoscaling’ tell the cluster to disable autoscaling

# this set the cluster size at its current default node pool size. And can still be manually updated.

gcloud container clusters update mycluster –no-enable-autoscaling –node-pool default-pool