slice-jump

This is a fairly minimal Vagrant project for spinning up a jump host in a openstack OpenStack environment. It's tested with Puppet's internal openstack OpenStack.

It also uses the Puppet provisioner to install Vagrant (and the vagrant-openstack-provider plugin) and other useful things such as:

mosh (mobile shell)
tmux
bind-utils (host, nslookup)
mtr

  Mac / Vagrant ----> OpenStack VM / Vagrant ----> OpenStack whatever project

I live in Australia and the network latency to the US means that vagrant-openstack operations are very slow. So one of the motivations for this project is to spin up a jumphost on OpenStack that I can then run vagrant on to control OpenStack on a much lower latency connection.

Using

There's a bunch of shell environment variables you'll need to set up first. There's an example set of them in openstack.rc. You may want to download your openstack.rc from the OpenStack UI, or copy and modify the example in this repo.

Source the rc file:

source openstack.rc

When prompted, provide your password.

In the Vagrantfile you'll see that the slicejump host is being created in network1. You may want to change this if you want your slicejump host in some special network you may have created.

Vagrant plugins:

Install vagrant plugin for openstack

vagrant plugin install vagrant-openstack-provider

Install Vagrant plugin for local host file updates if you prefer working with host names rather than IP address (optional)

vagrant plugin install vagrant-hostsupdater

Creation:

vagrant up --provider=openstack

Discover the SSH configuration (including floating IP of your new host):

vagrant ssh-config

Moshing in (alternative to ssh, uses udp):

ensure mosh is installed on your Mac or whatever your laptop / desktop is running - https://mosh.org/ (eg brew install mosh)
mosh --ssh "ssh -i ~/.ssh/slice.pem" ubuntu@IP_ADDR_OF_VM (replace IP_ADDR_OF_VM with the IP of the VM)
start up a tmux session tmux (or tmux a to reconnect to a previous tmux session)

What next?

Now you have the ability to use the vagrant-openstack plugin from a VM that's very close to the OpenStack API service, so you may want to take advantage of one of the following projects to assist with setting up an environment:

Puppet Debugging Kit - Puppet internal / Public
Oscar

Caveats

I set up a .tmux.conf file for user ubuntu which rebinds the tmux prefix to ctrl-a (default is ctrl-b in tmux, who knows why)
I'm using the puppet apply provisioner and everything is done in one file at puppet/environments/production/manifests/default.pp - buy me a Pirate Life IPA
Puppet modules have been vendored into puppet/environments/production/modules/ for reliability, speed, and simplicity

Why?

Using Vagrant to interact with OpenStack via the vagrant-openstack-provider plugin over the VPN from Australia is woefully slow. It can take minutes to get through the intial proceedings of authenticating to OpenStack and whatever else the openstack provider does before actually spinning up vms or anything else useful. I have a theory that this is due to the network latency between Australia and Portland, OR (around 200 ms on a good day) so I've created this project to encapsulate the task of setting up a host within OpenStack with which to control OpenStack for my other projects. It will probably also come in handy as a jump host to access my other OpenStack hosted vms from, etc.

How?

vagrant-openstack-provider creates a Linux vm (Ubuntu currently as it includes a package for mosh)

Vagrant's shell provisioner then ensures that:

Puppet's package repo for puppet-collections 1 is configured
puppet-agent package is installed
Probably 2 cores and 2 GB RAM is overkill for what this VM does.

Vagrant's puppet provisioner then takes over configuration and installs a bunch of packages, including Vagrant, tmux, etc.

To Do

Add gcc, make and other build tools?
Pass through OpenStack credentials somehow, maybe
See if this commit mades it through GitHub's outage