Well, we’re back from Belgium after another successful week of conferences. First we attended FOSDEM in Brussels, then we hustled down to Ghent for the CfgMgmtCamp speaker’s dinner. This year is special to me because it’s the first year that I’ve not gone to represent someone else. It was also special because the atmosphere was so positive. There were a ton of great talks, more sponsorship dollars, attendance was up, more people showed up for the optional third day add-on than ever before, and people responded really well to the talk lineup. Chef even showed up to try and rebuild their community engagement!
In other words, it felt a bit like a revival. But I’ll come back to that. (Or you can skip directly to it, just like online recipes!)
Main stage panel at FOSDEM
It started on Jan. 21, 10 days before FOSDEM. There were some last minute schedule changes that I won’t speculate about too much (they’re the same link, just two days apart) and I was invited to participate in a panel discussion about project forks, why they happen, what they can do to the community and so on. Read more about it and see the recording right on the FOSDEM schedule page.
What a fun experience! Dawn was the original architect of the Puppet community and was my boss for all of a hot minute when I moved onto the team. But somehow this is the first time I’ve shared a stage with her. Everyone on stage was super knowledgeable and I learned a ton. In particular, Stephen Walli explains that statistically two criteria predict the success of a community fork
- The community is put into existential crisis by actions of the original IP holder, and
- One or more respected non-employee members of the community step up to serve as anchors.
This bodes very well for the future of OpenVox.
A busy week in Ghent
The main stage lineup at CfgMgmtCamp was fascinating. First, Hazel Weakly talked about the humanity behind software, configuration, and infrastructure management. This is an interesting perspective to me. As I see it, humanity is the unsaid reason behind the devops movement in the first place but we’ve now shifted far away from it. It’s no surprise that Puppet’s slogan used to be the human-forward “freeing you to do what the robots can’t” and now it’s bland corporate scare-words “avoid business disruption and minimize downtime.”
The thematic arc of the morning took us through these ideals we held when starting the devops movement, to the nitty gritty of OpenTofu’s escape from the corporate thumbscrews and recapturing their own personal power, to the real value of a content ecosystem and musing on a better way to build sustainable ecosystems.
Of course, the rest of the conference was just as good. We had two full days of two solid Puppet / OpenVox tracks covering all kinds of topics, from the technical to the speculative to the damage control. On the main stage Adam talked about the hypergraph of the future. James kept you on your toes by throwing fire again. There were a couple of talks about compliance. Stahnke showed us how he put his EPEL and Puppet knowledge to work with NixOS.
But there was still a vibe running through. Florian talked about toxic corporate environments and Nick explained how “source available” is such a woefully inadequate substitute for open source and ways to better appreciate your contributors. A lot of talks hinted at disillusionment with corporatism and the current state of the world. Lunch conversations debated the merits of various open source licenses and whether or not the GPL would prevent the licensing smash-and-grab ploys that so many once-great companies are pulling today. I lost track of how many people thanked me for my “yes, I’m a dirty hippie” slide.
I even overheard someone at a table nearby talking about their challenges in forming a tech union. I didn’t catch who or where it was, so if this was you, I’d love to hear more about it.
And that’s not even mentioning the overwhelming interest in OpenVox. Our community dinner (traditionally the Puppet Community and now the Vox Pupuli Community) filled the first floor of the Royal India. This was a nice healthy increase from years past and brought together several generations of Puppet / OpenVox community collaborators.
Some people were there for their first time and some had been there since the beginning; back when CfgMgmtCamp was still called a Puppet Camp and none of us had been acquired by private equity yet. But we were all united by two common interests: 1) keeping a community alive despite the best efforts of Puppet’s current owner to repeatedly shoot its own feet, and 2) excellent Indian cuisine enjoyed with friends.
For context: the conference is two days long, but they book the venue for a third day to make space for mini-events to tag along and colocate. Puppet has traditionally used this for a Puppet Community Day; Ansible and Foreman have similar events.
The optional third day is where this enthusiasm really shone through though. Not only were there more people who stayed for the third day than in years past, but every single person left the Puppet Community Day to join the OpenVox working session breakout. And as a measure of their commitment, at 5pm there were still 40-50 people in the room still hard at it, which is unheard of. Most people trickle out to catch their trains during the day.
We did not get quite as much done as we’d hoped, but everyone had the opportunity to participate and we set the groundwork for a stable and healthy project. In short, we adopted a Fedora SIG style governance and decided to remain an independent project via OpenCollective for the time being. As we mature a bit, we’ll continue investigating the Linux Foundation and other options.
Notes for the entire third day, including the Puppet Community Day and the OpenVox working session breakout can be seen in our wiki, thanks to Gene Liverman and Alex Fisher.
Insights
I think the positive vibes were mainly the result of three different factors.
First, the obvious
I don’t need to spend much time on this. With the media coverage and the furor around the clumsy way Perforce is stumbling through its community changes and the blatant money-grab, it’s no surprise to see excitement in our community. We shall see how much of that excitement is lasting.
What’s old is new again
For years, it seemed inevitable; cloud was eating our lunch. Every workload was becoming cloud-native, whether or not it really fit there, and the idea of managing configuration of long-lived instances was beginning to feel quaint. But there was a lot of foreshadowing, especially for those of us who’d been around for a few boom-bust cycles. A lot of the same problems being solved, a lot of the same challenges being (re-)discovered, opaque builds, semi-reproducible artifacts. And so much of it is tied together with not much more than the shell scripts I used to build production chroots back in the day.
The expected savings from cloud-native workflows haven’t really materialized either. If your workload is suited for, designed for, and optimized for the cloud then it can often be cheaper. But costs that were trivial at testing or prototyping stages have a tendency to unexpectedly balloon out of control at scale.
It’s not that cloud is bad. We’re actually moving a lot of workflows (such as the OpenVox build pipelines) to containers and it’s drastically simplifying what we’re doing. But now that the craze is dying down, people are realizing that it’s not the be-all end-all that it was promised to be. There are a ton of workflows well-suited to cloud & containers, especially transient and stateless tasks. But there are also a lot of workloads that benefit from being on-prem, or on long-lived stateful instances, or just don’t need the layers and layers of complexity.
And for these jobs, good old fashioned configuration management is right here waiting with its predictable pricing. The pendulum is swinging back to a hybrid infrastructure and we’ll all benefit from playing in the same sandbox. This year we saw a resurgence in config management community investment. But we also had a full Kubernetes track. And I find that very promising.
The industry, the world, and everything
There’s a lot going on right now. There’s somehow a blustery orange man-baby in the American White House again. For some unexplained reason, that gives Elon Musk the ability to go Twitter on the American government and slash essential departments protecting everything from consumer rights to the environment. “AI” is unsustainably disrupting industries and exploding the wealth gap. A lot of companies & industries aren’t fully recovered from COVID yet.
And right in the middle of that are company after company turning to corporate vampires and sucking the open source communities that built them dry. Is it any wonder that people are activating?
Back in the early 2000’s, I regularly protested with the EFF on the Santa Monica Pier in support of all sorts of digital rights and privacy protections. Not all the bystanders knew who the EFF was or what we were talking about. But most did or were curious to hear about it.
I’m getting some of the same vibes now. People are realizing that they need to take their interests into their own hands again. VC money has been cheap and easy for a very long time. And along with the foosball tables, that meant quite a bit of budget flexibility. It didn’t matter so much if you spent a few days polishing up a patch to contribute upstream when the runway was so long.
That’s been on the downswing for a while, but this last year really brought everything to a head. The investors and board members have come calling and they want their pound of flesh. And when it comes down to it, your boss cares more about the bottom line and the market cap than they do about you or your silly open source ideals. The days of founding a startup with a pair of dice are done. No more Uber but for dogwalking or AirBnb but for retirement homes or Tinder but for telling jokes.
Now it’s time for solidarity. People are realizing that the company they once trusted doesn’t actually have their best interests in mind. They’re seeing through the permissive license and contributor license agreement blinders. And they’re viscerally coming to grips with the idea that very few businesses include “provide something of value” as an actual business goal.
And they’re starting to do something about it.