The Hardest, Most Rewarding Job I’ve Ever Had

Note: A couple of years ago, Engine Yard gave Mitchell Hashimoto an OSS grant to work on Vagrant. A few weeks ago, Vagrant 1.0 came out and Mitchell wrote a blog post about his journey and how he got to 1.0. With his permission, we’re reposting it here.

I just released Vagrant 1.0, exactly two years since I showed Vagrant to the world for the first time. I’ve made an official announcement but I think it is only appropriate to share a deeper, more personal story of the road travelled to reach this milestone, and what I’ve learned throughout the process. My goal in sharing is to give others an inside look at the guts of a successful open source project and perhaps offer a different point of view of the open source world.

Today is one of the proudest days of my life. I’ve released the first stable version of Vagrant, the software project I started with John Bender over two years ago while I was still in college. Vagrant is currently used by Mozilla, RackSpace, LivingSocial, Shopify, OpenStack, EventBrite, and many, many more. Both the project and the ideas behind the project have been far more successful than I could’ve ever dreamed of. But the road to this point is an interesting one, filled with highs and lows, and I’d like to share it with you.

The original idea for Vagrant came in 2009, when both John Bender and I were employed by Ruby development shops where it was routine to see a new project every 6 to 8 weeks. It was becoming increasingly frustrating to setup our development environments for every new project, which would be _slightly _different from previous projects. And it was absolutely infuriating when we had to go back to do maintenance on an old project since it was always a nightmare to get the environment back to that state. In the winter of 2009, I was frustrated enough to try to come up with a solution to this problem.

At this point in my life (and even still today), I viewed open source very romantically. People like Yehuda Katz and John Resig were my idols, because they did their work in the open passionately and successfully. I wanted nothing more than to find my own “jQuery” or “Ruby on Rails.” Basically: I was eager to start something which I thought would change the landscape of some field, just as jQuery and Rails did theirs. I don’t actually believe that Vagrant is as influential as jQuery or Ruby on Rails, but this gives you an idea of how I was thinking at the time.

In Janurary, 2010, I approached John Bender with some initial thoughts for what would eventually become Vagrant. John immediately saw value in the idea and offered to join in on the project, which I gladly accepted. We got started working immediately. It was January 21, 2010.

We both worked furiously, and within a week we had a working prototype that could bring up virtual machines we could SSH into. Below, you can see a screenshot from January 31, 2010, about a week after we started hacking, showing a functional version. Yes, Vagrant started life named “hobo.”

The initial development went fairly smoothly. Much of the essence of what Vagrant is today was molded during those initial weeks, and I’m happy to see that the ideas John and I had have been validated worldwide. Some are starting to show their age, but the fact that so many early decisions have lasted this long I think shows we had the right idea.

Fun fact: John came up with the entire idea of the “box” system about a week before the public release. Prior to that, I had planned to release Vagrant 0.1 only supporting a single Ubuntu image. The “box” system has been one of the most critical pieces in making Vagrant as successful as it is today, so my hat is off to John here.

While Vagrant was ready for release around mid-February, I was concerned that an open source project coming from two unknown/unproven developers would hurt our initial adoption. I decided that to be as successful as we could, we’d have to have amazing documentation and a mascot. Yes, a mascot was_critical_. I don’t know why, but I just find projects that have mascots to be more trustworthy. So, Vince was born (shown below), and I spent a week only working on documentation. I can’t stress how important this week was. Vagrant 0.1.0 was released on March 7, 2010 with a mostly positive response. There were a handful of individuals who immediately grasped onto the idea and began using Vagrant with their projects right away. Most of these individuals still use it today, and deserve recognition for being so brave to adopt a new technology so early. Some of these early adopters even border on fanatical, creating things like Vagrant pins they distribute anywhere they go. I love this: The first few months of Vagrant were reasonably uneventful. Besides the initial rush of early adopters, growth mostly stagnated, and each release of Vagrant was getting averaging around 100 downloads. While I loved the project and still whole-heartedly believed in it, seeing very little growth was hugely discouraging. At this point, I actually started to view Vagrant as a potential “failure.” Despite this, John and I and our respective companies used Vagrant every day and saw the value first hand. And I was still passionate about the project, so I decided to just keep going, believing that if this were truly a good idea, something good would happen.

And something good did happen, something great: Carl Lerche discovered Vagrant. At the time, Carl Lerche worked at Engine Yard and was a Ruby on Rails core developer. He also specifically pair programmed withYehuda Katz. Carl popped in and out of the Vagrant IRC channel for a few weeks, asking for help and offering ideas here and there, and even contributed a few times. After a few weeks, he private messaged me. I don’t remember the exact words he used, but it was something along the lines of “How would you feel if Engine Yard sponsored Vagrant?” I vividly remember shaking with excitement at this point, despite the uncertainty. Engine Yard had long been known for being huge supporters of open source in the Ruby community, and backed important projects such as Ruby on Rails, JRuby, Rubinius, and a few more. I saw this potential sponsorship opportunity as huge idea validation as well as an outlet to better spread the word about Vagrant.

On October 14, 2010, Engine Yard announced that I had joined their OSS grant program. The specifics of the deal are private, but the basic idea is that they would help me in any way possible, as long as it was reasonable. This was a really exciting day because it was the first time with Vagrant that I could say “Mom and Dad, look! See! I told you I’m not just playing on my computer.” At this point I was still in college, as well, just to put things in perspective.

The Engine Yard sponsorship changed everything. Just having Engine Yard supporting me spurred a huge interest in Vagrant, and blasted Vagrant into “small-time popularity:”

The personal horsepower I put behind Vagrant went up to over 9000 at this point. Since I was still in college, I was spending 8 or more hours per day on Vagrant. At the same time, I was sending speaking proposals_everywhere_ I could to educate people about Vagrant and try to gain some more interest in the project. I spoke at a handful of conferences, pushed many releases, and by March, 2011, the average page views per day on vagrantup.com had gone from around 200 to over 500. Success!

Unfortunately, this success came at a price: burnout. By March, although I refused to admit it for many more months to come, I was completely and utterly burnt out. This is clear to see from the release dates of various Vagrant versions: It took 6 months to release Vagrant 0.7.0 from Vagrant 0.6.0 (although there were various bug fix releases between). This was the lowest point in my personal involvement with the project. During these times, I would let bugs pile up to around 20 or 30 before triaging them all in one go. I’m not proud of this, but it was an important part of the history of Vagrant.

While I was burnt out and busy feeling sorry for myself, Vagrant only grew more and more popular. A great community built around Vagrant, a healthy set of plugins, and I gained a small fan club. It is the community that brought me back. I simply started getting more and more tweets, emails, etc. telling me how Vagrant had changed people’s lives, how they couldn’t imagine working before Vagrant, etc. I was flattered, and the praise was highly motivational. By the fall of 2011, I was working on Vagrant again, though not as much since I was out of college at that point and had a full time job.

In October, 2011, I travelled halfway around the world to Sweden for DevOpsDays, where I gave a talk on DevOps. It was here that something big happened: About 15 seconds into the talk, I introduced myself as the creator of Vagrant, in case anyone would recognize me that way. I thought maybe a handful would care, but instead the entire room, filled with around 200 people, erupted in applause, which you can hear in the video. This single act of kindness, again by the community, showed me just how much people cared about what I was doing, and motivated me even further. I consider this an extremely important moment in Vagrant history, and would be the first of many amazing events I’d witness in the following months.

Vagrant has been a full time job for the past 2 years. I work 8 hours at work not on Vagrant, and then spend at least 4 hours at home working on Vagrant, and typically also work on it on the weekends. I’m incredibly proud to finally ship a 1.0, and I’m proud of what the project has taught me and the community that has grown around it. I hope my story shows how much work, luck, and passion has gone into Vagrant. If I could go back in time, I wouldn’t change a thing, since as they say, “it’s all about the journey!”

I’m looking forward to see where this journey continues to take me.

And finally, last but not least, thank you so, so much to the Vagrant community and early supporters. Patrick Debois, Christian Trabold, Kieran Pilkington, and so, so many more: You make it a joy for me to work on Vagrant every day. Open source is all about the community. And, of course, thank you to Engine Yard for all their support, which continues to be critical in educating the world about Vagrant.