Skip to content

Latest commit

 

History

History
81 lines (55 loc) · 3.07 KB

README.md

File metadata and controls

81 lines (55 loc) · 3.07 KB
Raw

Provision a Simple Test Cluster

The Terraform config here can be used to provision a simple test cluster with kip.

VPC with KIP

Getting Started

You need:

  • an AWS account configured (check that e.g. aws iam get-user works)
  • Terraform >= 0.13.0
  • aws-cli
  • jq

Initialize Terraform and verify the configuration:

$ terraform init
[... lots of output ...]
$ terraform plan
[... lots of output ...]

Upload your SSH public key to AWS, and remember its name. Make sure the private key for this public key is your default SSH key: $HOME/.ssh/id_rsa. Alternatively you can use an SSH key pair generated from AWS. In this example we’ll name our key-pair “alice-key”. Come up with a unique cluster name like “alice-cluster”. Finally put these values into a tfvars file:

$ cat > myenv.tfvars <<EOF
ssh_key_name = "alice-key"
cluster_name = "alice-cluster"
EOF

There are other configuration variables available, there’s an example tfvars file env.tfvars.example. You can then apply your config:

terraform apply -var-file myenv.tfvars

This will create a new VPC and a one-node Kubernetes cluster in it with kip, and show the public IP address of the node when done:

[...]

Apply complete! Resources: 13 added, 0 changed, 0 destroyed.

Outputs:

node-ip = 34.201.59.101

You can now ssh into the instance using the username "ubuntu", and the ssh key you set in your environment file. (It takes a a minute or two for the instance to bootstrap). On the instance, you can use kubectl to interact with your new cluster:

$ kubectl get nodes
NAME                          STATUS   ROLES    AGE   VERSION
ip-10-0-26-113.ec2.internal   Ready    master   67s   v1.17.3
kip-provider-0                Ready    agent    13s   v1.14.0-vk-v0.0.1-125-g3b2cc98

If you haven't set an existing ssh key in your configuration, a new ssh key has been created. You can extract and use it via:

$ terraform show -json | jq -r '.values.root_module.resources | .[] | select(.address=="tls_private_key.ssh-key") | .values.private_key_pem' > /tmp/id_rsa
$ ssh -i /tmp/id_rsa ubuntu@<public IP of node>

AmazonSystemManagedInstanceCore permissions are attached to the node instance. You should be able to see node instance in AWS Console under Systems Manager -> Instances and Nodes -> Managed Instances. There you can start new SSH session. You can also connect to instance using AWS CLI: $ aws ssm start-session --target <instance ID in i-XXX format>

Run a Pod via Virtual Kubelet

The node taint in kip is disabled in the manifest, so Kubernetes will try to run all pods via the virtual node.

If you decide to enable the taint on the virtual node (via removing the --disable-taint command line flag), you will need to add a toleration and/or node selector for pods that are meant to run via kip:

spec:
  nodeSelector:
    type: virtual-kubelet
  tolerations:
  - key: virtual-kubelet.io/provider
    operator: Exists