FOSDEM '09 is a free and non-commercial event organized by the community, for the community. Its goal is to provide Free and Open Source developers a place to meet.


Interview: Jasper Capel

Jasper Capel and Robert Lazzurs will give a talk about Cobbler & Koan at FOSDEM 2009.

Could you briefly introduce yourself?

Sure! I'm fairly new to the open source world as a Contributor / Developer. I've started submitting patches to open source projects only about a year ago, as I then decided I didn't only want to use open source software; I also wanted to help make it better. Luckily, this goes hand in hand with my daytime job: I'm a technical consultant for Stone-IT, an open source consulting firm in The Netherlands.

What will your talk be about, exactly?

We have two talks, actually. At first, we were going to do a Cobbler talk/demo in the Fedora room, but since we've been asked to do a lecture in the Main Tracks, this allows us to split it. The Main Tracks lecture will be an overview of the Cobbler project. We'll explain how this project started, present the roadmap, explain some use cases and talk about integrating Cobbler in your environment.

What do you hope to accomplish by giving this talk? What do you expect?

We hope to get people interested in our project! Our current user base consists of mostly Red Hat, CentOS and Fedora users, so if we can reach users of other distributions, or *BSD, that would be great. If we can get some of them involved, even better. We're very keen on supporting more hardware and distributions. If we could, we'd probably install an OS yet to be invented on a coffee machine. Also we like to assimilate ideas from other folks who might have their own home grown provisioning environments. The idea behind Cobbler is that we can get everyone together to share their ideas and build up this one, really feature rich common codebase that has the features of /everyone's/ environment, so if someone starts up a new project they don't have to spend time building a new tool, they already have one they know available. It's very much everyone's project, and where we go is limited only by what ideas people have.

What was the motivation to start with the Cobbler project?

I can interpret your question in two ways: "Why did I start using Cobbler?" and "Why was the Cobbler project started". I'll just answer both, starting with the first: For one of my projects (developing a cloud / managed hosting environment), I was looking for a good way to provision many virtual servers at once. After some searching, Cobbler turned out to be that tool, as it was the only solution I could find that didn't involve creating system images. It gave us the ability to deploy a lot of systems that weren't 100% identical. For example: we'd make a base profile which all systems use, and then create sub-profiles for different use cases without much effort. The templating is done with Cheetah, so effectively we could just use Python in our kickstarts and snippets (small, re-usable pieces of code). All functionality I needed that wasn't in Cobbler, I just added.

Why was the Cobbler project started? I'll just quote the founder (Michael DeHaan) here. I've asked him the same question in preparation for FOSDEM, as I've only been using Cobbler for a little over a year myself ("we" in this context is Red Hat):

"Cobbler was started about 3 years ago when we were looking at gaps in OSS technology for systems management. There were plenty of good monitoring frameworks but we lacked a provisioning server, and none of our options addressed both bare metal installations and virtual installations at the same time. Other features that came along later (managed DHCP, managed DNS, yum mirroring, etc) were largely suggestions from users and we just kept adding to it as people had good ideas to speed up their deployments."

How big is the Cobbler developer community?

There are about 50 people who have contributed code to the project, and more that are active on the mailing-list and IRC (freenode, #cobbler) and are contributing their ideas.

Which distributions can we install with Cobbler?

Currently: Red Hat Enterprise Linux, CentOS and Fedora work really well and recently, support for Debian and Ubuntu has been added by Javier Palacios. Work has been done on supporting SuSE as well, deployments work now and we hope to get the import command working soon too. For what's to come: There are people working on adding Windows deployment support and we're speculating on Jumpstart support for Solaris. Folks interested in any of the above, or interested in adding support for other operating systems, are welcome to stop by the mailing list and share ideas.

How well does Cobbler integrate with other tools?

Cobbler integrates pretty well in both directions (information from Cobbler to other systems and vice-versa). We can manage DNS- and DHCP-servers, and support linking Cobbler to a configuration management system (Puppet, for example), so all of these can be managed from Cobbler.

There are also projects using Cobbler as their provisioning-backend. Examples are: Spacewalk and oVirt. Another cool example is Beaker, which is an upcoming Fedora QA project for test automation. Also lots of folks integrate Cobbler with their own in-house software over XMLRPC to make custom interfaces or connect it with other scripts. Cobbler comes with a powerful Python and XMLRPC API, so you can basically connect it to anything if you write the glue.

Which new features or improvements can we expect to appear in Cobbler in the next 2 to 3 years?

Two/three years is a long time, and judging by the current pace of development, we can get a *lot* done in that time. I (personally) hope we'll be able to deploy every major operating system by then.

Creative Commons License
This interview is licensed under a Creative Commons Attribution 2.0 Belgium License.