December 4, 2018
Thanks to the FreeBSD Foundation, I was able to attend MeetBSD, my first BSD conference and second open source conference ever. It was an amazing experience. As someone who lives in a remote rural area, the ability to directly interact with like minded people is invaluable. While learning is a huge part of conferences, the interaction with the community is vital too. The community is what makes the FreeBSD project what it is today, a thriving success, full of extremely motivated people. It was so much fun being able to see and converse with people I’ve read blog posts about or talked to online. I came to BSD for great performance and security, but I am staying for the community. My flight arrived at the San Jose International Airport just after noon on the 18th. It was a long day getting there, but it was worth it.
I arrived at the beautiful Intel campus bright and early at 7am on the 19th where I immediately met Michael Dexter. Michael and I had initially met in the spring at LinuxFest NW. I had such a fun time and he convinced me to come to a real tech conference. After registering and a very warm welcome by all the staff, we filed into the very nice auditorium and had some excellent breakfast pastries. While waiting in line I was able to jump into a conversation on the woes of some very nasty NFS bugs. I think it only took 30 seconds before I learned something new.
Ten a.m. rolled around and it was time for Kris Moore’s talk on using TruOS to boot-strap your FreeBSD projects. As someone who had previously been using TruOS as a desktop platform, it was quite interesting to see how it had evolved to something much more important. During the talk there was also an update on pkgbase, something me and many others would love to see come to FreeBSD. Up next was Ben Widawsky from Intel’s talk on how a big company like Intel uses and contributes to FreeBSD. The important takeaway for me was how Intel decides what to work on. Sorting features into different buckets and seeing which one’s lineup is a good way to see that everyone can benefit from their work. At lunch I was able to chat with a system administrator from the NASA Advanced Supercomputer about his experiences with big computing while using OpenBSD and Linux. After lunch it was time for a talk on NetBSD’s MKSANITIZER. A lot of this went over my head, but it was interesting to see how modern code sanitization works. After that we had some quick lightning talks including Allan Jude’s great piece on how he uses ZFS boot environments at scale. Boot environments really seem to give the sysadmin a lot of comfort knowing they have that working environment to fall back to if things go bad. Now it was time for Andrew Fengler’s talk on monitoring in FreeBSD and the value in collecting different metrics. Andrew and I had a great chat the following evening on software in military technology. The much anticipated ZFS panel answered some tough questions and gave us an update on upcoming features like device evacuation and native ZFS encryption. Allan was kind enough to fill in the blanks for me on why limiting the ARC size doesn’t solve the fairly heavy RAM requirement for ZFS. Limiting ARC size will alleviate issues with smaller files, however any files larger than the ARC limit will begin to cause issues. Then something quite funny happened, a mere mention of the Bloomberg/Supermicro VMC debacle and the entire room started their own discussion groups. Now it was dinner time. We all shuffled into the Intel café for some pizza. I was able to ask some more of my questions during the next few hours.
We were all back bright and early for Day 2. After picking up my copy of the ZFS book from Michael W. Lucas, it was time for his talk on why we all use BSD. BSD has always been about ingenuity and freedom, something outlined very well during the talk. Next, Dr. McKusick was kind enough to stop by and bless us with some hilarious stories of the very early days of BSD. Did you know that RMS modified his password utility to allow null passwords? I sure didn’t! Devin Teske graced us with her presence to show off her work with DTrace and dwatch with lots of pretty graphs on Grafana. It was very cool seeing all that data being put in Grafana. I guess I have to do the same thing for my home-lab now. After a quick lunch, we had a talk on programming and why it actually is important to use the right language. Security in FreeBSD is a priority, and a talk by Marius Zaborski on Capsicum showed us why. As someone who’s barely heard of Capsicum, this answered quite a few of my questions. Another panel, this time focusing on virtualization came next. FreeBSD’s bhyve hypervisor was a hot topic and it was very cool to have developers fielding questions in real time. As someone who is big on using FreeBSD in Windows environments, I was able to get some info on running FreeBSD on Hyper-V. During the breakout sessions I took the opportunity to talk to Chris Fisher about podcasting and drone flying. I am a drone pilot too and one advantage of my remote region is that I get to make drone videos of migrating whales. The final talk of the conference was Nick Principe’s talk on performance monitoring and why so many things “depend” on other things. He explained to me his title the previous evening. I greatly enjoyed his slide demoing the importance of identifying bottlenecks at various points in terms of the metrics you get out of monitoring. At our final evening and dinner, I was able to chat with Benedict Reuschling about some work I had done that would hopefully modernize some FreeBSD documentation.
Without the gracious help of the FreeBSD Foundation it would not have been possible to attend MeetBSD. I met brilliant people whom I admire, that were willing to answer a handful of my questions, with many more to ask next time. Best of all I learned the ways I can get involved with the FreeBSD project to make sure new people like me can have an amazing experience like the one I have had.
– contributed by Conor Beh