Skip to content

Latest commit

 

History

History
181 lines (121 loc) · 6.95 KB

File metadata and controls

181 lines (121 loc) · 6.95 KB

WARNING WARNING WARNING WARNING WARNING

PLEASE NOTE: This document applies to the HEAD of the source tree

If you are using a released version of Kubernetes, you should refer to the docs that go with that version.

The latest release of this document can be found [here](http://releases.k8s.io/release-1.1/docs/getting-started-guides/dcos.md).

Documentation for other releases can be found at releases.k8s.io.

Getting started with Kubernetes on DCOS

This guide will walk you through installing Kubernetes-Mesos on Datacenter Operating System (DCOS) with the DCOS CLI and operating Kubernetes with the DCOS Kubectl plugin.

Table of Contents

About Kubernetes on DCOS

DCOS is system software that manages computer cluster hardware and software resources and provides common services for distributed applications. Among other services, it provides Apache Mesos as its cluster kernel and Marathon as its init system. With DCOS CLI, Mesos frameworks like Kubernetes-Mesos can be installed with a single command.

Another feature of the DCOS CLI is that it allows plugins like the DCOS Kubectl plugin. This allows for easy access to a version-compatible Kubectl without having to manually download or install.

Further information about the benefits of installing Kubernetes on DCOS can be found in the Kubernetes-Mesos documentation.

For more details about the Kubernetes DCOS packaging, see the Kubernetes-Mesos project.

Since Kubernetes-Mesos is still alpha, it is a good idea to familiarize yourself with the current known issues which may limit or modify the behavior of Kubernetes on DCOS.

If you have problems completing the steps below, please file an issue against the kubernetes-mesos project.

Resources

Explore the following resources for more information about Kubernetes, Kubernetes on Mesos/DCOS, and DCOS itself.

Prerequisites

Install

  1. Configure and validate the Mesosphere Multiverse as a package source repository
$ dcos config prepend package.sources https://github.com/mesosphere/multiverse/archive/version-1.x.zip
$ dcos package update --validate
  1. Install etcd

By default, the Kubernetes DCOS package starts a single-node etcd. In order to avoid state loss in the event of Kubernetes component container failure, install an HA etcd-mesos cluster on DCOS.

$ dcos package install etcd
  1. Verify that etcd is installed and healthy

The etcd cluster takes a short while to deploy. Verify that /etcd is healthy before going on to the next step.

$ dcos marathon app list
ID           MEM  CPUS  TASKS  HEALTH  DEPLOYMENT  CONTAINER  CMD
/etcd        128  0.2    1/1    1/1       ---        DOCKER   None
  1. Create Kubernetes installation configuration

Configure Kubernetes to use the HA etcd installed on DCOS.

$ cat >/tmp/options.json <<EOF
{
  "kubernetes": {
    "etcd-mesos-framework-name": "etcd"
  }
}
EOF
  1. Install Kubernetes
$ dcos package install --options=/tmp/options.json kubernetes
  1. Verify that Kubernetes is installed and healthy

The Kubernetes cluster takes a short while to deploy. Verify that /kubernetes is healthy before going on to the next step.

$ dcos marathon app list
ID           MEM  CPUS  TASKS  HEALTH  DEPLOYMENT  CONTAINER  CMD
/etcd        128  0.2    1/1    1/1       ---        DOCKER   None
/kubernetes  768   1     1/1    1/1       ---        DOCKER   None
  1. Verify that Kube-DNS & Kube-UI are deployed, running, and ready
$ dcos kubectl get pods --namespace=kube-system
NAME                READY     STATUS    RESTARTS   AGE
kube-dns-v8-tjxk9   4/4       Running   0          1m
kube-ui-v2-tjq7b    1/1       Running   0          1m

Names and ages may vary.

Now that Kubernetes is installed on DCOS, you may wish to explore the Kubernetes Examples or the Kubernetes User Guide.

Uninstall

  1. Stop and delete all replication controllers and pods in each namespace:

Before uninstalling Kubernetes, destroy all the pods and replication controllers. The uninstall process will try to do this itself, but by default it times out quickly and may leave your cluster in a dirty state.

$ dcos kubectl delete rc,pods --all --namespace=default
$ dcos kubectl delete rc,pods --all --namespace=kube-system
  1. Validate that all pods have been deleted
$ dcos kubectl get pods --all-namespaces
  1. Uninstall Kubernetes
$ dcos package uninstall kubernetes

Analytics