[OpenStack-Infra] Vancouver Summit v2 Recap

Clark Boylan cboylan at sapwetik.org
Tue May 29 21:05:44 UTC 2018


Hello everyone,

Not everyone was able to attend the summit and even for those of us that did we were not able to attend all the sessions related to Infra. I'm hoping this thread can help us all make sense of the busy week we just had and give us a general sense of the direction for the next 6 months or so of work. I'll try to give a high level overview of the topics I was involved in, feel free to ask for or add specifics if you think it will be helpful.

The first day of the summit was largely Zuul's day in the spotlight. Corvus had an excellent keynote to talk about Zuul. After keynotes were a variety of forum sessions that related to Infra team work. 

Storyboard migrations came up. These migrations seem to be going reasonably well and we have identified some specific areas to improve like search performance and web client deployment improvements.

In the machine learning for CI session we've identified that there are individuals interested in doing machine learning on the logs we produce. The big barrier there is getting the logs to a location with the appropriate hardware and tools for teaching machines what to look for. To start we've pointed people at grabbing the logs from our existing logs server to start understanding what we collect and doing testing, but we'll need to publish somewhere with a bit more hardware power (GPUs) for longer term work.

Doug Hellmann proposed a plan for deprecating python2 within OpenStack. We'll need to make sure we can accommodate the needs of that deprecation process on the Infra testing platform. We should also keep in mind we ourselves will need to move off of python2 and head towards python3.

Days two and three I spent all of my time in the OpenDev conference within a conference. Pleia2 did an awesome demo of deploying to "production" starting with nothing but a cloud. "Free software needs free tools" by Benjamin Mako Hill was probably my favorite talk of the entire event. Explains why the Infra team is determined to use free tools to build software.

The OpenDev working sessions were great venues to interact between open source communities and other organizations. We share many of the same concerns and have to address many similar problems. A lot of time was spent feeling out a common vocabulary and constraining the problem space of CI and CD. The attendees were happy to use https://openci.io to continue these conversations longer term, hope to see you there.

OpenDev also had hands on workshops. Shout out to pabelanger and mnaser who ran an excellent zuul workshop.

The fourth and last day of the summit was a busy one for me. Dmsimard and I gave an Infra team onboarding session which had great attendance. After that I was on a security panel. Then we had a conversation with the Kata team to talk about how we can start running Zuul jobs against Kata. Plan there seems to be get it started on one of their lower volume repos and start running a job they can interact with.

My summit largely ended with the Infra project update that I gave. The last sessions I participated in was somewhat of an extension of the Kata discussion earlier where we talked about cross community governance between OSF projects. Big outcome of this was that we need to communicate but at the same time don't be surprised if communicating is difficult. Many of us come from different places and don't have the shared OpenStack background to work off of.

This email got long very fast and I feel like I've barely scratched the surface. Happy to dig in where people are interested and to hear from others on what their experiences and takeaways were.

Thank you,
Clark



More information about the OpenStack-Infra mailing list