devops4me/kops repository overview

⁠Kops | kubectl | AWS Cli

With docker run we'll provision both a simple .k8s.local and a production grade Kubernetes cluster. The [Dockeerfile] takes care of installing kops, kubectl and the AWS Cli.

⁠The Todo List

• create a k8s4me user in Dockerfile
• create a docker volume and copy the public key into it
• when running create cluster mount volume and link public key file to /var/opt/k8s4me/.ssh/id_rsa.pub
• check that the volume now has a .kube folder in it
• maybe use safe to suckup the ~/.kube/config file
• clone git app deployment project
• run kubectl commands by mounting .kube/config and mounting the git deployment yamls
• possibly add ~/.kube/config file to kubernetes secrets on Jenkins stack
• now build jobs (after pushing docker image) can use kubectl container to deploy them into cluster

⁠Part 1 | Create a .k8s.local Kubernetes cluster

⁠Step 1a | Checklist

• clone this project | git clone https://github.com/devops4me/docker-kops
• step into the project | cd docker-kops
• create an S3 bucket to manage the kubernetes cluster state (kops.kubernetes.cluster.state)
• create a comma separated list of availability zones
• setup an AWS IAM user with the below 5 IAM permissiones
• create a RSA public/private keypair and place the public key in k8s-admin-key.pub

AmazonEC2FullAccess
AmazonRoute53FullAccess
AmazonS3FullAccess
IAMFullAccess
AmazonVPCFullAccess

⁠Step 1b | Create a .k8s.local Kubernetes Cluster

Run the docker build after placing the RSA public key (2048 or 4096 bits) in the k8s-admin-key.pub file. Next set the S3 bucket name, the cluster name, the default region, IAM user keys and the availability zones. Then execute the docker run statement to create the kubernetes cluster.

docker build --no-cache -t img.kops .
docker run --interactive \
           --tty \
	   --rm \
	   --name vm.kops \
	   --env KOPS_STATE_STORE=s3://kops.kubernetes.cluster.state \
	   --env KOPS_CLUSTER_NAME=lab.dublin.k8s.local \
	   --env NAME=lab.dublin.k8s.local \
	   --env AWS_DEFAULT_REGION=`safe print region.key` \
	   --env AWS_ACCESS_KEY_ID=`safe print @access.key` \
	   --env AWS_SECRET_ACCESS_KEY=`safe print @secret.key` \
	   -v $PWD:/root \
	   img.kops kops create cluster --zones "eu-west-1a,eu-west-1b,eu-west-1c" --master-count 3 --master-size=t2.micro --node-count 3 --node-size=t2.micro --yes

⁠Step 1c | SSH onto a Master Node

Find the IP address or public DNS name of any master node and ssh into it like so.

ssh admin@<<ec2-public-dns-name>> -i /path/to/private-key.pem

⁠Step 1d | Use kubectl to manage the cluster

If you run kubectl too early you'll get a Unable to connect to the server: EOF error message.

docker run --rm --volume $PWD:/root img.kops kubectl get nodes -o wide

⁠Step 1e | Delete the Kubernetes cluster

We can delete the entire VPC wrapped Kubernetes cluster by specifying the cluster name, state s3 bucket store name and IAM user's region and access keys.

docker run --interactive \
           --tty \
	   --rm \
	   --name vm.kops \
	   --env KOPS_STATE_STORE=s3://kops.kubernetes.cluster.state \
	   --env KOPS_CLUSTER_NAME=lab.dublin.k8s.local \
	   --env NAME=lab.dublin.k8s.local \
	   --env AWS_DEFAULT_REGION=`safe print region.key` \
	   --env AWS_ACCESS_KEY_ID=`safe print @access.key` \
	   --env AWS_SECRET_ACCESS_KEY=`safe print @secret.key` \
	   img.kops kops delete cluster --yes

⁠Checklist | b4 creating a production Kubernetes cluster

• clone this project | git clone https://github.com/devops4me/docker-kops
• step into the project | cd docker-kops
• register a domain name and have the hosted zone ID handy
• use dig NS devopswiki.co.uk to check access to the domain name
• create an S3 bucket to manage the kubernetes cluster state (kops.kubernetes.cluster.state)
• decide on an availability zone
• setup AWS credentials and add to the appropriate ENV variables

dig NS devopswiki.co.uk
safe login <<book.name>>
safe open kubernetes lab.dublin
docker build -t img.kops .
docker build --no-cache -t img.kops .
docker run --interactive \
           --tty \
	   --rm \
	   --name vm.kops \
	   --env KOPS_STATE_STORE=s3://kops.kubernetes.cluster.state \
	   --env KOPS_CLUSTER_NAME=lab.dublin.k8s.local \
	   --env NAME=lab.dublin.k8s.local \
	   --env AWS_DEFAULT_REGION=`safe print region.key` \
	   --env AWS_ACCESS_KEY_ID=`safe print @access.key` \
	   --env AWS_SECRET_ACCESS_KEY=`safe print @secret.key` \
	   img.kops