Kevin's musings

I decided to take a slightly different way home. In part because I couldn’t find a flight from DRS -> AMS that would get me there in time for the AMS->PDX leg and partly just because it would be fun:

I took the train from dresden to frankfort and then on to Utrecht. It was about a 10 hour journey, but it was nice. I went with Patrick and Randy and we had a bunch of discussions and watched the landscape go by. In Utrecht I got a hotel for overnight and we went to a great little resturant along the canal. Great place. My only regret it that it got too late for me to see Patrick’s place, so I just went to bed and got up the next morning for my flight.

On checking in I saw that I had a lovely SSSS on my pass, so I got to get extra searched at the gate, but it wasn’t too bad, and it let me board before I would have normally been able to. Then the flight back and the drive home were mostly uneventfull and I managed to get in around 2pm.

Overall this was a great flock, I had a lovely time and all the organizers did a awesome job!

Day 4 jumped right into sessions:

Protecting the batcave against the Joker: Fedora infra hardening workshop Michael Scherer. Some interesting vulnerabilities that misc found (but did tell us about in advance), and how they were found and some ideas to help mitigate such things moving forward. The big one was a predictable git checkout in tmp, so we could add noexec to tmp and/or do seperate tmp for all users. Running ansible-lint also is something we should really look at doing.

After that I again jumped to hallway track, and then a team lunch, which I was late getting back from.

After running back late from lunch I got to the Fedora Infrastructure hacking and planning and hacking workshop session. I wrote up a bunch of areas we wanted to talk about or work on in the coming year. There really wasn’t too much discussion on most of them, they were all things we knew we wanted to do it’s just a matter of doing them. Neal Gompa did a live demo of the current pagure 5.0 themes and appearence. It looks nice!

The list was:

• new cloud – this is just very very close, so we decided to finish it asap.
• project tracking – I will setup a test for this once pagure 5.0 is out.
• replace nagios – many people suggested we look at zabbix again.
• CAIAPI/noggin – asked folks to file requests for features
• Repospanner – very close to existing, just need to finish it off.
• fedora messaging – abompard gave us a overview of where we are and we discussed corner cases like FMN and datanommer.
• sigul 1.0 – pretty much done we just need to have a flag day to update everything in step (since it’s not compatible with 0.9)
• copr – we discussed plans to move parts of copr into main infra. Possiblly frontend in openshift and dist git and keysign replaced by sigul.

We had a number more we didn’t get to because we ran out of time:

• python3 all the things
• more openshift
• more ci stuff
• javascript frameworks

Next up was my Infrastructure Fix Fest. I was hoping we would get folks to come to us with all the small bugs they hit that we just haven’t had time to fix up and we could fix them right then in the session. Unfortunately, I think because it was the last session of the day and later in the week, many people went back to their rooms to rest for the evening activity or went to the hallway track to talk to people. We did get a few things fixed… I managed to close about 5 infra tickets, track down and fix a issue with kerneltest application that Justin wanted me to fix and discuss a few others. I think next year if we do this again I would ask people for items before hand and we can use it as a chance to show how to fix things.

The evening entertainment was a ‘roller coaster’ restaurant. You order from a tablet and they send your beer or food down on a set of rails. It was pretty amusing. There was a lot of sci-fi clash (the servers were wearing star trek the next generation outfits, but there were star wars characters all around too). At one point a bottle of beer broke, but no one was hurt and they got it cleaned up pretty fast. Overall a lot of good company and amusement.

After a nice sleep and breakfast things started out with announcements then the state of Fedora talk from the FPL, Matthew Miller.

Summary: doing pretty well, but we should try and get the numbers of users estimates more accurate, and keep doing the good work on all the things we have in the pipeline for the next few years.

Fedora Objective leads then gave updates on their objectives.

• Matthias Clasen talked about all the good stuff in Fedora Workstation over the last few releases
• Langdon White updated us on Modularity
• Stephen Gallagher talked about Fedora Server and directions for upcoming work
• Dusty Mabe talked about Feodora CoreOS and Atomic host. We should see a combined setup for f30.
• Peter Robinson talked about Internet of Things and plans for it coming up.

Overall a great bit of info on where we are and what we had done this last year. There seemed to be more excitement in the air than last year.

In the middle of the above reporting we took a coffee break. I won’t mention them again, but I’d like to note here that they worked really really well. There was an area setup with coffee and pastries and some tables with no chairs and everyone could mingle around and talk. It was a really great way to have a “hall way” track without missing any talks. Since it was all in the same area you could often find whoever you wanted to talk with, or wait for others to come talk to you. Kudos to the organizers!

Lunch was in the hotel resturant, buffet style. Was pretty good overall and was nice to avoid travel time, etc.

After lunch, I went to the “RHEL, Fedora and CentOS: Solving the penrose triangle” talk by Josh Boyer and Brenden Conaboy. It’s very true that there is not a normal direct upstream -> downstream flow to the relationship between these three, and they had some interesting ideas on how we could make the feedback better and understand it more.

Next was the Fedora Infrastructure onboarding talk by Ricky Elrod. This is a topic that is near and dear to me, so it was great to see the ideas and also to see the interaction from the audience. There were a bunch of questions and ideas. I hope we can implement some of them soon and see if they help folks out or not.

After that, I went to the session on improving compose times with Mohan Boddu and Lubomir Seldar. This one had a bunch of good ideas on how we can get our compose processes down in time from the current 8-9hours to much less. There was a lot of audience participation here. Great to see!

The next slot there were several things I wanted to see, but I ended up in the hallway track talking to lots of people, so that was all good too.

I think the talks this year were all recorded so they should appear soon I hope and folks can see them for themselves.