devstack-vagrant

This is an attempt to build an easy to use tool to bring up a 2 node devstack environment for local testing using Vagrant + Puppet.

It is almost fully generic, but still hard codes a few things about my environment for lack of a way to figure out how to do this completely generically (puppet templates currently hate me under vagrant).

This will build a vagrant cluster that is L2 bridged to the interface that you specify in config.yaml. All devstack guests (2nd level) will also be L2 bridged to that network as well. That means that once you bring up this environment you will be able to ssh stack@api (or whatever your hostname is) from any machines on your network.

Vagrant Setup

Install vagrant & virtual box

Configure a base ~/.vagrant.d/Vagrantfile to set your VM size. If you have enough horsepower you should make the file something like:

VAGRANTFILE_API_VERSION = "2"

Vagrant.configure(VAGRANTFILE_API_VERSION) do |config|
    config.vm.provider :virtualbox do |vb|

         # Use VBoxManage to customize the VM. For example to change memory:
         vb.customize ["modifyvm", :id, "--memory", "8192"]
         vb.customize ["modifyvm", :id, "--cpus", "4"]
     end
end

You can probably get away with less cpus, and 4096 MB of memory, but the above is recommended size.

If the used hostnames in the config.yaml file (variable hostname_manager and hostname_compute) are not resolvable you have to install the vagrant-hostmanager plugin (vagrant plugin install vagrant-hostmanager).

If the nodes are still not able to communicate to each other even after installing the vagrant-hostnamanger plugin (for example you get errors about the compute node not being able to communicate to cinder c-api during the vagrant up phase), set the variable use_ip_resolver in the config.yaml file to true, in order to obtain the correct nodes ip.

Local Setup

Copy config.yaml.sample to config.yaml and provide the hostnames you want, and password, and sshkey for the stack user.

Then run vagrant up.

On a 32 GB Ram, 4 core i7 haswell, on an SSD, with Fios, this takes 25 - 30 minutes. So it's not quick. However it is repeatable.

If you want to speed-up the process, install the vagrant-cachier plugin in order to let vagrant cache files, such as apt packages, with:

vagrant plugin install vagrant-cachier

What you should get

A 2 node devstack that includes cirros mini cloud image populated in glance. You can get other images population such as fedora 20, ubuntu 12.04, and ubuntu 14.04, just with a small addtion to extra_images part in config.yaml.sample.

Default security group with ssh and ping opened up.

Installation of the stack user ssh key as the default keypair.

Enable additional services

The devstack environment created by this Vagrantfile includes just the basic services to get started with OpenStack. If you want to try more services, you can enable them on the manager node through the config.yaml file.

For example if you want to enable Swift, you can add the following line to your config.yaml:

manager_extra_services: s-proxy s-object s-container s-account