Can my Vagrant VMs use manifests with storeconfigs without a Puppet master?

Question

I'm trying to set up a multi-VM Vagrant environment that spins up an OpenStack controller and a few OpenStack compute nodes and then provisions a VM or two on the compute nodes.

Yes, I'm talking about VMs running on VMs. It sounds a little crazy, but this multi-VM Vagrant environment has already been created at https://github.com/lorin/openstack-ansible and it works perfectly, as I describe at http://wiki.greptilian.com/openstack

I can only draw inspiration from that GitHub repo and its corresponding tech talk, however, because it uses Ansible as the Vagrant provisioner while I have a requirement to use Puppet.

I'd like to use the official Puppet Labs module for OpenStack at https://github.com/puppetlabs/puppetlabs-openstack but it uses storeconfigs, so I get errors like this because my Vagrantfile doesn't include a VM to serve as a Puppet master:

warning: You cannot collect exported resources without storeconfigs being set; the collection will be ignored on line 142 in file /tmp/vagrant-puppet/modules-0/nova/manifests/init.pp

Resource type anchor doesn't exist at /tmp/vagrant-puppet/modules-0/nova/manifests/db/mysql.pp:18 on node controller.example.com.

I suppose I could tweak my Vagrantfile to spin up a Puppet master along with the OpenStack hosts, but I'm not sure how I'd do that and it seems to introduce extra complexity into the Vagrant environment.

I'm wondering if can do this in with "masterless" Puppet instead. A post at http://semicomplete.com/presentations/puppet-at-loggly/puppet-at-loggly.pdf.html suggests it's possible, saying, "puppet –environment prerun manifests/prerun.pp ... makes storeconfigs work ... puppet –storeconfigs manifests/site.pp ... This is the main puppet run" but I'm confused about the implementation details.

Can anyone point me to a Vagrant repo that runs "masterless" Puppet but uses storeconfigs?

Answer 1

You'll need to configure your storeconfigs with a DB that all vagrant VMs can reach. Loggly used AmazonRDS, but you can use other DBs as puppet docs show. Assuming you have a DB that all VMs can reach and you run puppet with storeconfigs option and you have the correct DB connection info configured in puppet, you should be good.