Over 400 developers gathered in Dublin, Ireland at the most recent Project Teams Gathering (PTG) to collaborate on their respective OpenStack projects. There was a veritable zoo present (each project has an animal mascot) as over 40 teams worked tirelessly for five days despite a record winter storm, giving birth to the hashtag #Snowpenstack .
Missed this PTG? Save the date for the next one: it will take place in Denver the week of September 10-14, 2018.
Rich Bowen, open source evangelist at Red Hat, set up an interview station at the Dublin PTG. Here are some of his highlights.
The OpenStack community gathered in Dublin to plan the next release – Rocky – of the OpenStack software. I took advantage of this rare opportunity to set up my tripod and interview some of the community leaders about what’s new in Pike, what’s coming in Rocky and their thoughts about the OpenStack ecosystem.
These interviews have slowly been making it on to my YouTube channel and there are still more to come.
I’ll start with my final interview. On the last day of the event, Thierry Carrez, the OpenStack release manager, sat down with me to discuss the PTG, the future of PTGs, and the OpenStack community as a whole. This was a great way to wrap up the week since his knowledge of the OpenStack community is deep and goes back to the beginning, giving him useful insights on the future.
The takeaway here is that we’re still evolving how the community works, because everything about the community, the software, and the users of the software, is changing all the time. We will have the next PTG to play for the Stein release, but beyond that, we are still listening to the community for direction.
I had numerous other conversations with various projects — I encourage you to have a look at the full play list (https://bit.ly/2uPkHaC) ) Here are a few highlights to get you started.
We all rely heavily on Zuul. It’s the magical box that tests all the things and how they interact with all the other things, every time we change a line of code. This massively complex interaction can cause a change in one place to break something way over there that we didn’t think was related and Zuul ensures that this doesn’t happen.
I chatted with Jim Blair and Jesse Keating from the Zuul team to talk about what Zuul actually is and what they’re working on.
There’s also a second interview, coming soon, about the way that Zuul is being used in communities other than OpenStack to do the same kind of interaction testing.
Special Interest Groups
SIGs are a relatively new thing in OpenStack. These used to be called working groups, but working groups are typically tasked with solving a particular problem. The SIG concept has grown out of that, but is more about topic-based discussions that spans multiple projects.
The best way to illustrate this idea is to look at two specific SIGs and what they’re working on.
I spoke with Michael McCune from the OpenStack API SIG, and Stig Telfer from the OpenStack Scientific SIG. The API SIG works across all projects to ensure consistent and predictable APIs. The Scientific SIG works with projects that have any high performance computing (HPC) or scientific computing applications, to ensure that they’re working together effectively, and implementing best practices from that larger community.
The Cyborg project is one that I find fascinating, because it’s less about what functionality is exposed to end-users and more about implementing the interfaces with the hardware that it runs on. It’s one of the projects that many users will likely be completely unaware of, but that they’re relying on every day.
Cyborg provides a framework for interacting with the many acceleration resources that OpenStack has to run on, including crypto cards, GPUs, FPGAs and so on.
As such, this team has to work with numerous hardware vendors and help them implement their drivers, get their patches into the upstream, and document how to use them. This kind of cross-industry communication is something that OpenStack has to be good at, because of the many places that it gets used.
Finally, I’ll mention one project that’s sort of outside of the family. Kayobe is a project that grew out of a need to provide a custom installation of OpenStack for the scientific community and has grown into a more general purpose installation/management project, hoping to address some of what they feel are the shortcomings of TripleO.
They’re also working to become a top level project at OpenStack, and the conversation around the requirements for that was very interesting to me. OpenStack wants to welcome everyone, but, at the same time, has a process in place to ensure that if something gets to call itself “OpenStack” it has to be of a certain quality.
These are just a few of the 30 interviews that I conducted on-site at the PTG, as well as the ones that I’ve done online since that event. I’m still working to get them all edited and posted, so please subscribe to my YouTube channel – http://youtube.com/rdocommunity – so that you catch the rest of them as they come online.
Check out the next Project Teams Gathering (PTG) for details on the upcoming event.