Hi all, We tentatively plan to meet next Tuesday January 24th at 4-5pm PST / 7-8PM EST. We will have a zoom link sent out on this thread shortly.
As a part of the meeting, we will discuss our meeting cadence going forward but here is a tentative agenda. This meeting may be split into two sessions depending on how much we’re able to get through in one hour. Agenda: 1. Discuss what to do with Akka and if we need to continue outreach to Lightbend. Apache Pekko is under way, but first release is not yet available. 2. Discuss if we’re ready to perform a new major release, 2.0.0. 3. The removal of zookeeper dependency for the invokers. The removal of kafka from execution path can be a part of this discussion as well even if they are two distinct units of work. 4. Discuss if there’s anything else we can do to improve performance of new scheduler this year. 1. Have the decision maker consider action concurrency 2. Have two average durations one for cold start init and one for a normal warm execution. Things along those lines. 5. Discuss if we can add container concurrency to other runtimes 6. Triage any other desirable user facing features. I have the action level container concurrency limit feature in progress but haven’t had bandwidth to finish it. There’s the in progress action versioning pr that’s been almost done for a couple years. Things along those lines. 7. Do a pass through core repo updating all libraries to deal with transitive security vulns. See you there, Brendan Doyle On Wed, Dec 14, 2022 at 6:05 PM Dominic Kim <style9...@gmail.com> wrote: > Thanks, Brendan for initiating this discussion. > > Regarding the 2.0.0 release, I believe we can initiate it by the > beginning of the next year, and also we can consider enabling the scheduler > by default. > Second, it is indeed required to optimize the annoying scheduler build > process and it would be also helpful to make OW more easily run on M1 mac. > > They are great agendas and +1 for all of them. > > IMHO, one of the most important matters we need to discuss is not only many > committers but also most PMCs are less active these days. > We need to review the current community status and reorganize members. > We are now definitely short-handed. > > > Best regards > Dominic > > > > > > > 2022년 12월 15일 (목) 오전 3:04, Brendan Doyle <bdoyle0...@gmail.com>님이 작성: > > > Hello all, > > > > I'm hoping that after the holidays, we can bring back the technical > > interchange. Potentially we can start with a quarterly endeavor. Over the > > last year and some change, we've made some dramatic re-architecturing to > > the core project through the new scheduler contributed by Naver. I > believe > > that the scheduler is now nearing production readiness. > > > > It also feels like there is a renewed interest in the project in 2022 > with > > many new people trying to deploy and play around with Openwhisk through > the > > deploy kube repo, presumably for academic research. Any way to get these > > folks more involved and welcomed to the community would be great. > > > > For January the main things I would like to tackle agenda wise are: > > > > 1. What is next for the new scheduler architecture. > > > > - When do we cut a 2.0.0 release and start removing code / cleaning up > the > > old architecture. Can we start a regular release cadence moving forward? > > - What optimizations can we tackle for the new scheduler in 2023? > > - How do we promote the existence of this major re-architecting which > > dramatically improves the core performance? > > > > 2. Dealing with dependency upgrades. > > > > - It's been a long time since many libraries have been upgraded and > there's > > likely security updates that need to be applied. > > - Discuss the current status of Akka licensing and the Pekko open source > > fork. > > > > 3. What can / should we accomplish in 2023. > > > > - What features can we focus on in 2023 now that the major effort of the > > new scheduler is mostly done. I have some ideas I posted in slack, but > > would love to hear from community members on this as well. > > - Are there any other re-architectures that we can feasibly tackle (i.e. > > removal of kafka off the critical path) or new technologies that have > come > > into the field since 2018 that could dramatically improve Openwhisk that > we > > should analyze adopting. > > > > Please respond if this is something you would like to or could attend so > I > > can get a gauge on interest and start organizing a date and time. Even > just > > a couple people I think will make the meeting worthwhile so that we can > > come up with a high level roadmap for the next year. > > > > Happy Holidays, > > Brendan Doyle > > > > [image: Mailtrack] > < > https://mailtrack.io?utm_source=gmail&utm_medium=signature&utm_campaign=signaturevirality11&; > > > Sender > notified by > Mailtrack > < > https://mailtrack.io?utm_source=gmail&utm_medium=signature&utm_campaign=signaturevirality11&; > > > 22. > 12. 15. 오전 11:00:19 >
