Introduction

This repository is intended for demo-ing the manual install of kubernetes's components on both master and worker nodes.

It should be able to get you to a working single master (insecure) kubernetes setup on a set of VMs

@startuml
database etcd [
  etcd 
]

package "master-node" {
  [api-server] -> etcd

  [kubelet] --> [api-server] : watch
  [kubelet] --> [container-runtime] : run & watch
  
  [scheduler] --> [api-server] : watch
  [scheduler] --> [api-server] : apply

  [controller-manager] --> [api-server] : watch
  [controller-manager] --> [api-server] : apply
}

package "worker-node-1" {
  [kubelet ] --> [api-server] : watch
  [kubelet ] --> [container-runtime ] : run & watch

}

package "worker-node-2" {
  [kubelet  ] --> [api-server] : watch
  [kubelet  ] --> [container-runtime  ] : run & watch
}

@enduml

prerequisites

• vagrant
• the scp vagrant plugin : vagrant plugin install vagrant-scp
• the GNU parallel CLI
• jq

setup

• run vagrant up to start the vms. This will create a master node and 2 worker nodes on your host's network

• run ./scripts/show_cluster_config | tee cluster.config

• copy the cluster configuration to the nodes:

./scripts/copy_file_to_nodes cluster.config

• install the jq CLI on the nodes so they can read the config

./scripts/run_script_on_nodes install_jq_cli

• setup a container runtime

./scripts/run_script_on_nodes install_container_runtime

• download kubernetes

./scripts/download_kubernetes_binaries $(cat cluster.config | jq -r ".kubernetes_version") ./kubernetes

• download etcd

./scripts/download_etcd_binaries $(cat cluster.config | jq -r ".etcd3_version") ./etcd3

• copy kubelet & kube-proxy on the worker nodes

./scripts/copy_file_to_nodes ./kubernetes/workers worker

• copy kubelet, proxy, apiserver, scheduler and native controllers binaries to the master nodes

./scripts/copy_file_to_nodes ./etcd3 master