Docker Kubernetes get started quickly

Today, Kubernetes supports installation in a variety of environments, including on-premises hosts (Fedora), cloud services (Google GAE, AWS, and so on). However, the quickest way to experience Kubernetes is clearly to start the process locally through Docker.

The following illustration shows the topology of a set of Kubernetes deployed quickly in a single node using Docker.

Kubernetes relies on the Etcd service to maintain the state of all primary nodes.

Start the Etcd service

docker run --net=host -d gcr.io/google_containers/etcd:2.0.9 /usr/local/bin/etcd --addr=127.0.0.1:4001 --bind-addr=0.0.0.0:4001 --data-dir=/var/etcd/data

Start the primary node

Start the kubelet.

docker run --net=host -d -v /var/run/docker.sock:/var/run/docker.sock  gcr.io/google_containers/hyperkube:v0.17.0 /hyperkube kubelet --api_servers=http://localhost:8080 --v=2 --address=0.0.0.0 --enable_server --hostname_override=127.0.0.1 --config=/etc/kubernetes/manifests

Start the service agent

docker run -d --net=host --privileged gcr.io/google_containers/hyperkube:v0.17.0 /hyperkube proxy --master=http://127.0.0.1:8080 --v=2

The test status

Access the 8080 port locally and get the following results:

$ curl 127.0.0.1:8080
{
  "paths": [
    "/api",
    "/api/v1beta1",
    "/api/v1beta2",
    "/api/v1beta3",
    "/healthz",
    "/healthz/ping",
    "/logs/",
    "/metrics",
    "/static/",
    "/swagger-ui/",
    "/swaggerapi/",
    "/validate",
    "/version"
  ]
}

View the service

After all services are started, look at the Docker containers that are actually running locally, and there are several.

CONTAINER ID        IMAGE                                        COMMAND                CREATED             STATUS              PORTS               NAMES
ee054db2516c        gcr.io/google_containers/hyperkube:v0.17.0   "/hyperkube schedule   2 days ago          Up 1 days                               k8s_scheduler.509f29c9_k8s-master-127.0.0.1_default_9941e5170b4365bd4aa91f122ba0c061_e97037f5
3b0f28de07a2        gcr.io/google_containers/hyperkube:v0.17.0   "/hyperkube apiserve   2 days ago          Up 1 days                               k8s_apiserver.245e44fa_k8s-master-127.0.0.1_default_9941e5170b4365bd4aa91f122ba0c061_6ab5c23d
2eaa44ecdd8e        gcr.io/google_containers/hyperkube:v0.17.0   "/hyperkube controll   2 days ago          Up 1 days                               k8s_controller-manager.33f83d43_k8s-master-127.0.0.1_default_9941e5170b4365bd4aa91f122ba0c061_1a60106f
30aa7163cbef        gcr.io/google_containers/hyperkube:v0.17.0   "/hyperkube proxy --   2 days ago          Up 1 days                               jolly_davinci
a2f282976d91        gcr.io/google_containers/pause:0.8.0         "/pause"               2 days ago          Up 2 days                               k8s_POD.e4cc795_k8s-master-127.0.0.1_default_9941e5170b4365bd4aa91f122ba0c061_e8085b1f
c060c52acc36        gcr.io/google_containers/hyperkube:v0.17.0   "/hyperkube kubelet    2 days ago          Up 1 days                               serene_nobel
cc3cd263c581        gcr.io/google_containers/etcd:2.0.9          "/usr/local/bin/etcd   2 days ago          Up 1 days                               happy_turing

These services fall into three categories: primary node services, work node services, and other services.

The primary node service

• Apiserver is the external interface of the entire system, providing reSTful for clients and other components to call;
• Scheduler is responsible for scheduling resources and assigning a pod to a node;
• Controller-manager is responsible for managing the controller, including endpoint-controller (refresh service and pod association information) and replication-controller (maintaining the values that a pod is copied to configure).

The work node service

• Kubelet is the agent of the work node to perform the operation, is responsible for the specific container lifecycle management, according to the information obtained from the database to manage the container, and report the pod operation status, etc.
• proxy provides access to services on the pod.

Other services

• Etcd is a storage database for all states;
• gcr.io/google_containers/pause:0.8.0 is a test image that automatically pulls down after Kubernetes starts.