This Quickstart assumes that you have installed all of the dependencies.
The KDK works on many OS's, and may run as a docker container or virtual machine.
- On Mac and Linux, the preferred route is to run the KDK as a docker image.
- On Windows, the preferred route is to run the KDK as a Vagrant virtual machine (Hyper-V and Virtualbox are both supported).
curl -sSL https://raw.githubusercontent.com/cisco-sso/kdk/master/files/install | bash
kdk init && kdk ssh
The Vagrant Virtualbox image works on Mac, using the same instructions as the Windows version. The only reason one may choose to use this run method is if support for IPV6 networking is required. Docker for Mac does not support IPV6.
git clone [email protected]:cisco-sso/kdk.git # or https://github.com/cisco-sso/kdk.git
cd kdk
# Edit Vagrantfile: You may want to tune memory, network settings, or host-mounted directories.
vagrant up # Starts the KDK
vagrant ssh -- -A -D 8000 # Connect to the KDK (-A ssh-agent forwarding, -D socks proxy forwarding)
# Use the KDK
vagrant destroy
Please use Windows10 powershell for installation.
Set-ExecutionPolicy Bypass -Scope Process -Force
iex ((New-Object System.Net.WebClient).DownloadString('https://raw.githubusercontent.com/cisco-sso/kdk/master/files/install.ps1'))
kdk init ; kdk ssh
NOTE: After installation, Windows CMD prompt will work. The KDK has not been tested with Cygwin, Mingw, or Windows Subsystem for Linux.
Detailed installation instructions of the KDK along with all of its dependencies are found here:
Getting your workstation setup to work with Kubernetes clusters may require the install and configuration of quite a few tools. Do it inconsistently among your team, and your automation and workflows may not work properly for everyone. Even if it works on your machine because the latest code you've written requires the latest version of kubectl
and a new installation of jq
, your teammates Billy on Windows and Jane on Mac are busy filing bugs against your latest PR because they haven't received the memo about updating their toolchains.
We've created the open-source Kubernetes Development Kit (KDK) in order to solve this problem. The KDK is a docker container or a vagrant virtual machine, which may be deployed on Mac, Windows, and Linux. It is a Linux-based environment which has over 30+ tools pre-installed and pre-configured. If your team uses the KDK, then you are guaranteed to have a similar development and operations environment, with the same tools, at the same versions.
A sampling of tools include: docker, kubectl, helm, helmfile, kops, kubetail, docker-compose, terraform, ansible, minio-cli, aws-cli, gcloud, drone-cli, direnv, golang, git, hub, jsonnet, vi/vim, emacs, python 2/3, pipenv, pyenv, jq, zsh, helm-s3, kafkacat, dig, ssh-keygen, dep, gomplate, minikube, neutronclient, openstackclient, supernova, virtualenv, yq, colordiff, nmap, screen, tmux, sshuttle, yadm, and many others.
The KDK may make your life easier if you often:
- Operate Kubernetes clusters.
- Deploy Kubernetes clusters to various clouds including AWS and GKE.
- Develop and deploy Helm Charts.
- Develop docker containers.
- Create or re-create the config
kdk init
- Connect or reconnect to the KDK (will pull and start container if necessary)
kdk ssh
- Destroy the KDK
kdk destroy
- Update the KDK (binary, config, and container)
kdk update
The KDK is meant to be ephemeral. You should be able to kdk destroy && kdk ssh
whenever you need to reset your environment. Resetting should be done often, because over time your environment will diverge from original state as you use it.
Here are a few approaches for saving state in between resets.
The KDK default dotfiles includes a default .bash_profile
that will search for additional bash profiles in a few pre-defined locations. If files in any of these locations exist, they will be sourced automatically.
$HOME/.bash_profile_private
$HOME/.config/kdk/.bash_profile_private
/keybase/private/<user-keybase-id>/.bash_profile_private
- If the user has installed Keybase
Thus, one may customize their own private settings by creating any of the files above by host-mounting directories into the KDK when prompted during kdk init
. This method may be used to set environment variables as well as create entire dotfiles, such as ~/.aws/credentials
and ~/.aws/config
. See here for an example.
Upon KDK init, you will be prompted to mount additional directories from your host system into the KDK system. Typically this is used to mount code directories from the host machine to the KDK, but it can also be used to mount configuration directories.
Here's an example of mounting the ~/.aws
directory from an OSX machine to a location within the KDK.
Would you like to mount additional docker host directories into the KDK? [y/n] y
Please enter the docker host source directory (e.g. /Users/<username>/Projects) /Users/mcboats/.aws
INFO[0022] Entered host source directory mount /Users/mcboats/.aws
Please enter the docker container target directory (e.g. /home/<username>/Projects) /home/mcboats/.aws
INFO[0026] Entered container target directory mount /home/mcboats/.aws
If you are using OSX, then you may use ssh-agent to automatically forward your SSH keys into the KDK. This will allow you to access SSH resources (such as git cloning from Github) without physically copying your keys into the KDK machine, which lowers security. OSX automatically starts ssh-agent automatically. To load your keys into the agent, add your default keys with ssh-add
. From inside of the kdk, you may list which keys you have loaded with ssh-add -l
If you have your own yadm dotfiles repository, you may kdk init
with the option:
--dotfiles-repo string KDK Dotfiles Repo (default "https://github.com/cisco-sso/yadm-dotfiles.git")
NOTE: There are many configuration options available in kdk init
.See kdk init --help
for details
You might have a need to run multiple KDK containers. The KDK CLI can do that!
- Create a new KDK config
- NOTE: name parameter must be unique (no other container can have this name)
kdk init --name kdk1
- Connect to
kdk1
container
kdk ssh --name kdk1