Multi Node Kubernetes developer environment - using VBox/Vagrant/CoreOS

 Features of an Effective a Developer Environment

• Convenience and Productive
  
  • Fast/Easy to install - everthing for development included by default
  • Fast/Easy to start/stop/update/teardown
  • Ability to reproduce and share environments
  • Easy to manipulate multiple instances/environments and operate simultaneously
    
    • (golden/dev/test/experimental), compare/contrast/diff environments
    • experiment with federation and multi-site deployments in DevEnv
  • Export/Import/shareable instances - reproducable in many contexts
    
    • Share with others, publish, reproducable environment as part of bug reports, ...
• Reliability and Isolation
  
  • Instances isolated from each other and the developers system and from on-line requirements
  • Encapsulated - full environment enclused in the developer environment
    
    • Run multiple clusters in a laptop in an airplane disconnected from internet
  • Intentional upgrade process - avoid breaking auto-update
  •  (not automated),  encapsulated (minimal dependencies on external system/tools)

• Accuracy and Portability
  
  • Represents the 'production' implementation of the system as closely as possible
  • Run in all popular developer laptops/system/OS combinations
    • Windows, Linux, Mac, ...
  • Developments in DevEnv are directly applicable to production systems
  • Support development at all-levels as closely as possible
    • system-level, operations-level, apps, user-facing, and development at all levels
• Extensibility - Ability to extend and customize the platform as well as the developer environment
  • I.e.  no specific tool or model lock-in, freedom to change
  • Network - emulate multiple physical networks, ability to change k8s network plugins
  • Distributed - Emulate multi-site, multi-instance in dev
    • Federation, Distributed/federated/multi-site/edge-oriented development and testing

5 minutes to a running multi-node cluster that can run in your laptop.  VBox machines running CoreOS on host-local networking, not exposed to internet.  Notes below for other options.

Here is the recipe, adapted from the official instructions here: https://coreos.com/kubernetes/docs/latest/kubernetes-on-vagrant.html

1. Install VirtualBox
2. Install Vagrant
   1. Check vagrant: "vagrant box list"
3. Install kubectl
   1. curl -O https://storage.googleapis.com/kubernetes-release/release/v1.5.2/bin/linux/amd64/kubectl
   2. install in /usr/local/bin or your favorite spot in your PATH
4. git clone https://github.com/coreos/coreos-kubernetes.git
5. cd multinode/vagrant
6. vagrant box add http://alpha.release.core-os.net/amd64-usr/current/coreos_production_virtualbox.json
   1. will download CoreOS for virtualbox
7. cp config.rb.sample to config.rb and edit:
   1. Pick the channel and node sizes and counts according to your laptop resources (primarily memory), here's one example:
      

      $update_channel="alpha"

      $controller_count=1

      $controller_vm_memory=512

      $worker_count=2

      $worker_vm_memory=1024

      $etcd_count=1

      $etcd_vm_memory=512

8. Good to check your networking for collisions on unrouted address ranges that you'll encounter at work, home, etc.
   1. The default VM network specs are in the Vagrantfile:  172.17.4/24

9. vagrant up --provider virtualbox

10. export KUBECONFIG="${KUBECONFIG}:$(pwd)/kubeconfig"

    1. Alt:   Add the config to any existing ~/.kube/config if you are working with multiple clusters and or namespaces (kubectl set-cluster, kubectl set-context, etc.)

11. Should be operational, some useful commands (use in the multinode/vagrant path)

    1. vagrant ssh c1 (or the other names) to get into coreos

    2. vagrant status

    3. vagrant suspend/resume

       1. Once in a while net doesn't come back, and kubectl says:  

          The connection to the server 172.17.4.101:443 was refused - did you specify the right host or port?

       2. In this case you can vagrant halt, vagrant up

    4. vagrant destroy (obliterates)

       1. Just vagrant up and restart afresh

Enjoy,

D

Notes:

1. Can be used for volume/fs work by adding VBox disks to nodes.   Kubernetes doesn't have a VBox/vagrant volume driver (yet), so you get raw disk on nodes like we have in OpenStack.

2. Can be used for ingress/network by adding additional interface to the node to be LB.

3. It can be left running, or vagrant suspend/resume can freeze/thaw the whole cluster in a few seconds if you want it toiling around in the background.  Can also vagrant halt/up to shutdown/reboot and cluster state (etcd) is preserved.