Kubernetes Steering Committee, 9 May 2018

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Steering Committee 20180509

Description

Chat:
00:10:13 briangrant: https://github.com/kubernetes/kubernetes/pull/62088
00:10:33 briangrant: https://github.com/kubernetes/kubernetes/issues/44505
00:10:50 briangrant: https://github.com/kubernetes/steering/issues/21
00:13:52 Joe Beda: Eclipse foundation: https://www.eclipse.org/org/documents/Eclipse_Policy_and_Procedure_for_3rd_Party_Dependencies_Final.pdf
00:14:02 Joe Beda: https://www.eclipse.org/legal/licenses.php
00:21:30 Joe Beda: https://github.com/kubernetes/community/pull/1994
00:22:38 Quinton Hoole: Here’s the stuff I was referring to: https://compliance.linuxfoundation.org/references/tools
00:28:20 Timothy St. Clair: +1
00:47:45 Joe Beda: Feb 14 meeting in notes
00:59:25 Michelle Noorali: enjoyyy

A

B

B

So I don't see anything listed on your topics should I deny added talked about. There's.

C

Agenda items, but they were from last time because last time we didn't have a meeting, so you want me just carry forward the topics from last time: yeah, okay,.

B

B

D

C

Know if we'll have critical mass, though.

E

A

Am very surprised, given that this is day three of build house.

B

Build going, or is this your excuse to not be there right now there are no important important committee business.

A

Brennan, don't you have a talk today around this time.

B

Turns out we're all on stage right now,.

F

We're in has a.

A

Transparency, a.

F

Look at kubernetes from the inside, oh.

B

Nice stormtrooper helmet, they.

F

Attend what stormtrooper element.

E

And we're starting to actually get critical man, a suite.

B

Right, I volunteered to be note-taker for this I think we're about as corn as we're going to get. If somebody else wants to be chosen.

E

I'm happy to try and keep us on task here.

E

We go all right, so the first topic is CN. Cf governance board meeting items from michelle licensing black was allow versus disallow list proposal will send out slides.

A

Yeah I didn't I didn't do that.

A

Voted on, basically, the CN CF has to approve, like all of the licenses, for projects in the CN CF and their dependencies, so the project stopped dependencies and their licenses should also be in compliance, and so, instead of having to do that manually one by one, they just want like a blanket approval and just allow disallow lists that got voted on I think there was a gay all around so no no objections so that I got passed, and so that's not to say like if you're, not on the white list, that you know you know you can't be like an approved license.

A

It just means you have to go through the manual process. Yeah.

G

I think the bigger problem is the lack of continuous, automated enforcement. They run projects through false ology or something like that on a by request basis, which usually they only do when projects are first ducted into CN CF. So we actually sick testing folks we're working on automating the license verification. It turns out to be kind of tricky because most other than the apache license licenses, don't say what they are, so you actually have to pattern match the text as one of the.

F

Few people who refused the licenses finding that problem and can now map pattern match most of my site. Yeah.

G

Well, automating, it turned out they're like I, don't know eighty different versions or something just you know the handful that are actually supported. We also have a bunch of third-party code that doesn't also import the license files, so that needs to be fixed. We have there's an issue filed on steering from Christiana syczyk I believe with the list of things that were believed to not actually meet the requirements.

G

One of them, the rain, litter library, has been fixed, I think twice. It was fixed and unfixed agree. Sixth, but mostly I mean to remove from the project and then it got reacted. Yes,.

F

E

F

Rate limiter one: yes, wasn't that LGPL with the static linking exception, we have a bunch of those and they were deemed okay, yeah.

H

They were a long ago.

F

H

Remember we did that.

H

A

E

Too, but they're they're hard to pattern match because.

H

It's some like random string. They do, but to put it in the front of the LGPL yeah. So.

B

Just I know there are a couple of us here who were also at that open source leadership summit and I felt like automated automated compliance. Verification was repeated topic and many talks there, many of which I did not attend. Does anybody know if the talks were recorded in any way, shape or form, or could we reach out to that community of folks to see if we're not duplicating effort, I.

A

Don't know if the talks all of the talks were court I, don't actually think they were yeah.

G

I think they're not.

B

But I mean that's: definitely it is a wheel. Many people have worked on I hope we're not reinventing it. Like I'm hoping we've actually looked at some of the prior art out there. That's all I have I. Personally, as the same testing person haven't seen a proposal come across our mailing list about this. That's why everything up document prior art and stuff well.

G

Yeah I was went through our contributor experience.

A

I'm sorry, there are action items that we want to take from this conversation or take well.

G

We should probably so the P are from Ryan is still open and we, you know, I, don't know if there's a sergeant put in fossa for.

B

Context, what is this PR from Ryan I kind of missed that.

G

It is 62 Oh a date. Let me find the right window to put it in the resume.

B

Cat is that way, but just like was this something that addresses one of the licensing concerns. It's.

G

Is the thing that tries to pattern match all the acceptable licenses? Okay, I see so.

B

This is say, testings effort got it.

G

Yeah and whether it's suggesting or a contributor experience or whatever is- and the original issue was CC to contributor experience for whatever that's worth but yeah Ryan usually is more involved in testing.

E

Okay, so two things is that number one this feels like this should be something that should be a pre submit check that we can actually make on every single.

G

Yeah I don't think it needs to be pre semitic and just run as a test like we're, trying to not actually add more resubmit scripts like implement.

H

It as a unit test as a brightness.

G

H

Could work I think, except for it isn't the license? Extraction doesn't require good as well and.

E

The thing that I'm, the thing that I'm worried about is that, like, as we look at associated projects and stuff like that, we have the CLA bod that actually improves stuff. There should be a license checker. That's part of that also right, because otherwise this stuff will just drift right and.

F

Honestly, like the communities kubernetes repo, we have machinery in place that scrubs through all of the vendor directory fines license files. If they're not checked in it actually goes to the git, repo and crawls up the tree until it finds a license or copyright file copies all of those into a single place, so that when somebody sends a change to vendors, there's one file that we can that I personally flag. If that file ever changes, it needs special special review.

F

That was a requirement by Google legal early on and in fact, I think that licenses file is distributed as part of our tarball, and that was also a requirement from Google legal I. Don't know if CN CF holds a different position than Google legal on that, but I know that none of the other repos do that. Yeah.

E

I'm just saying this seems like a CNC F wide thing, not a kubernetes kubernetes thing agree.

G

Yeah I, don't remember if we previously filed a service desk issue on this I may have, but we certainly could I. Basically, after Chris filed an issue, I I basically said this is impossible. Without automated enforcement down about.

B

Other projects would handle this half or half of the project single there's.

G

B

G

Have not handled this, not ciencia projects anyway,.

B

If there is still a bucket of money, this certainly seems a bucket of money that the four discrete tasks that the scenes promote to contractors. This seems like one of them. Yes,.

G

There is I mean it looked like something was gonna happen, so that may have first all stepper and effort, but it looks like Ryan's. Er has been stalled for quite a while.

A

Is there something I should take back to the CAF and ask or a request.

G

Well, we should probably just have someone on the steering committee owning it and making up and making sure something gets done, and then you know we can decide whether it finished Ryan's PR work to file a service desk request. Ask the students you have to to do it. I, don't know what Ryan's PR.

F

Is but if you send me the link, I guess it's.

G

In the notes, okay,.

F

Since I've been dealing so much with it, I guess it makes sense that I would follow through.

A

Thank you, yeah.

G

I totally agree that the scenes yep has money and would be willing of money into it. We told them we weren't gonna, do solve it ourselves. So much.

A

Money y'all so much money. That's.

G

A

E

G

E

The CNC F has more money than it knows what to do with at this point. Yes,.

A

E

Yet please buy a diamond sponsorship to cube con or America steering.

H

Committee has position.

A

Okay, so I just you know: I wanted to bring up the white loss whitelist blacklist thing because it seemed like that's something from the meeting. That's relevant or that's something that you might want to know about going forward. I was thinking like. Would it be helpful if I sent out a summary of stuff that happened, that the governing board meeting on our steering private list? Is that something okay, I.

G

Thought that would be useful because they don't send out any official notes. You know I've asked them many times for this. It's I think it's not cool for it to be so Paik, but yeah yeah.

A

I guess on the steering private list, the only I can like send out the sent out some notes and relevant things and relevant slides, um for example, like they're. The majority of the last meeting was like mark. It was just going through the marking plan for the next year. So if we had suggestions, as the steering committee on you know, asks or requests and feedback, there, then I think they appreciate it. I I gave them my thoughts, but I'm sure that everybody else has has thoughts too.

A

So I'll send out the relevant information and start doing summaries. We'll see how that goes. So I'm.

E

The licensing thing sorry, sorry Michelle. It sounds great idea.

E

Does anybody have marketing stuff before you bring us back to Licensing sorry, we.

A

Can continue that, let.

G

You do my thoughts here. I apologize, I just want to say, I've seen the marketing duck feet of my other ciencia Pat and it's a huge step forward compared to last year. I. Don't.

E

Know I said that there was one slide that had solution three times in one sentence. It's.

G

Not perfect for sure, but it's so much better than what they were discussing last year and.

A

They're so open to feedback, so shirts.

E

So I'm the licensing thing I'm just trying to think like what concretely can we do to actually sort of you know, I agree that automation is really this. The the right way to go here. Could we do a better job of evangelizing to folks that they have to take this stuff seriously? Could we write it down someplace in terms of like where we say how to add a dependency and make sure that it's the list is linked there? Could we, when we find that bad dependencies get added?

E

You know, take somebody you know aside and say: hey you didn't follow this process. You got to clean up this mess. I mean like I, feel like there's things we can do to actually start like because I'm looking at like say, I just looked at the Eclipse Foundation as an example, and it doesn't what they look. They have automation, but they definitely have some clear guidelines that are on a website that then obviously their maintainer czar, aware of so.

B

I, do you think contributor experience might be the right place to talk about this and make sure that it's in the contributor guide and developer guide that they're working on going forward? I've attended the past couple of the meetings and I, don't recall hearing about this license thing so I can make sure that it's raised. You know this.

G

Is an old, fairly old issue like it's not a recent thing, it's been ongoing for eons yeah.

F

And concretely, I guess what I'd like to see is if CN CF hat holds a legal position on what makes good licensing, and maybe even that you know we go back to our definition of associated rico's and say: not only do you have to sign the CN CF CLA, but you need to ensure that you, following these licensing and here's a script if you're using go programs if you're writing go code, here's the script that you can run as part of your CI, which will verify all the licenses of your dependencies trans. Happily,.

E

F

We strongly urge any project that considers itself to be associated to run the script yeah.

G

The Associated repos just require the apache license yeah, but yes, I agree with what Joe said, but is far relying on human effort, where the project of our scale is just proven impossible, like you have way too much churn in people involved in, and the guidelines also change over time and people never go reread them so I I, don't disagree.

E

But it's not top of mind for folks. It's not something that people are thinking about, as they add dependencies so I feel like we can do some education here that this is something that folks should be aware of that it's not a free-for-all in terms of adding whatever code they find random out on the Internet to the code base right. Yes,.

G

I mean we definitely should have documentation, and we should have a clear policy and we will probably require email to kubernetes death or something to make people aware of it at least initially, but that'll just be a point in time like the people who join the week after that won't know about it.

B

So I can take an action item, Carol bump this invisibility and priority for the Trinity experience. I'm sure this is something that Tim pepper, probably has some significant experience with and could ride a little bit more. You know if you're talking about educating people they're, definitely the groups that it kind of does that and we can make sure that broadcasting the kubernetes def may be talking about community meeting making sure it's documented. That's that's all they're. Do we do.

E

We do like what is it at all.

G

Nobody have more than twice as many contributors now is when I earn. This is when I did that, let's.

B

Just do the slack game day, just let him know five minutes.

E

All right so about time box is spent 20 minutes on that bullet point anything else on the governance board, meeting stuff, Michelle, sorry.

I

Just just one last comment on the licensing stuff, I believe the CNC F actually has quite active stuff around automated license management and I. Don't know the details, but I remember going to a speech once where they described like annotating licenses with metadata and automatically you know verifying those things might be worth just having a quick look at what they can do there for us. Well,.

G

I already asked Chris about this and he didn't mention anything other than fossa. So, okay.

I

I mean if we have covered that base. That's fine I had thought that they actually had plans to provide like material support in the space of automated license, verification, stuff.

G

If they do, Chris doesn't know about it, I can.

A

Double-Check I think that that's the direction that they're headed in so I can double-check.

E

Okay, so the next topic is various governance, clarifications PR, so this was Tim and Phil Oh, which LEDs it best.

C

That's him so the there was a couple of there were a couple of folks who had questions comments, complaints, concerns about the process for filing their charter and I know that Phil had done a PR in the meantime. In the comments of what the process is and I think, I just think we need to start evangelizing and making sure that it's clear for folks who are submitting charters, what the expectations are and how the process should work. Yeah.

E

Now is actually my poll request which I started before I went on vacation. Then it got hung up, I I, think there's another issue here as where, as we do this stuff in the open, that's one thing, but we need to find a way to close on this stuff amongst ourselves because, like otherwise it's so hard to get this stuff in well.

G

So one thing we talked about doing is getting half a dozen charters and looking at all of them and deciding what we like and don't like and then agreeing on some good examples for other people to follow before we really press on asking the other six to move forward with their charters. There are half a dozen ish that are in progress now, so we could actually do that. I provided feedback to the gaps, one that put some effort into the Sagarika texture one I read: this is a scalability one.

G

At some point, there's also one from sig node, which Derek created I, believe contributor experience actually already merged. There's.

E

All right so so it seems like we want to beta-test the process, but even a beta test, the process. We need some sort of process that is at least a spit ball with this stuff more than we have now, because we have the example charters, which is one thing, but then there's also the process of like how do you go about getting reviews? When is it done? Who do you talk to? How do you close this out in timely manner and I feel like we don't have a lot of clarity around that.

G

Isn't particularly good, oh it's I mean we did have the spreadsheet, where we assigned every sig a couple of steering committee members to help them with the process. So I've been doing, you know doing a sign up for, but.

E

But like I guess, the question is: is that like? How do we get to the point like? Where is there a vote? Is there consensus amongst the steering committee members when is a vote called like I? Think we just need to have some of that stuff nailed down, because you know I think these things get started and then there's not. There ends up being a like. Well what next and even amongst the steering committee I, don't think we're all on the same page in terms of what next to actually get these things closed and merged.

B

My suggestion would be that we hold ourselves to have deadlines. I, think that we I feel like we were discussing the same thing when we came up with. Let's look at the six examples and we had supposedly sent ourselves. The deadline of next meeting will actually present the six charters, so we collectively get an agreement on what a good charter looks like right, and then we never get that I feel like we kind of owe it to ourselves to circulate through the group. What actually looks good so?

B

Could we hold ourselves to that this time, so.

A

Like say, the deadlines like in two weeks, the next meeting like do, we just come back to this meeting and like we vote and like if its majority yeah, then it's in. Why.

C

Don't we do a a subgroup again like we've done four other things, so subgroup can like take initiative and go forth, and you know at least the TLDR for the rest of the folks to process.

B

Yeah, like I'm, not so much talking about the meta process that Jose can Jose seems to be constrained with here. I wasn't sure if we have reached consensus on what like a good quality versus poor quality Charter is, and that's what I thought you're trying to accomplish here. So my suggestion be of the six charters that have been put forth using a spreadsheet where we have steering committees assigned to those SIG's and their charters. Those folks should help hold themselves accountable to actually, you know highlight: what's working, what's not working in those charters actually.

G

Got yeah I also found that trying to write the Charter from the template was instructive because there were things that became clear, that the template didn't cover it all or didn't cover in the right amount of detail, or was worded oddly things like that, so I want to feed those back into the template. Although I haven't done that yet so I.

B

Don't know my father was if this is the sort of thing where we want to come to consensus and we don't necessarily want to show that sausage being made. We can share our comments privately and then sort of reflect on this at the next meeting or we can share it on an ongoing basis on the public mailing list.

B

But if you like, this is one of those things we could really do offline with via mailing list threads and then just sort of circle back at the next meeting and then having gotten that consensus at the next meeting, we can move forward with the rest and I think Joe. Perhaps if you want to put together the framework for that meta process and like should they contact? Who are the reviewers? We, you know which one we can have a strong men at by next meeting as well. Yeah.

A

Is that public or sorry? Actually, one thing before that Aaron? That makes a lot of sense? I did not know that that was the process that you were talking. Yes, I love that, let's do that and by let's do that. I mean I vote for like let's talk about this privately on the steering private list and kind of keep our comments to ourselves.

A

The other thing was actually I forgot. My other comment now darn I.

B

Will just I usually get up at even when it's about like backroom conversations? The reason I want to have a private is I, don't want to like call specific people out in public. That's all when it comes to like well. Those that's few now, especially.

A

Because it's not their fault, because this is all like not like super solid. Yet sorry or.

G

If they're questions about scope of a cig or things like that, that might be controversial, so we might want to discuss that in private first. If there are major questions about that.

E

All right so I think the first thing we need to do is we need to actually communicate to the rest of the community that we consider the chartering process to be in alpha, and so we're doing it with a couple of early examples so that we can work out bugs because I think right now, there's a lot of confusion in terms of what should SIG's be doing and giving them the freedom to say. Hey we're still working this stuff out. Does that seem reasonable.

E

I think I asked that at a community meeting, but an email might reach more people yeah, and we just need to find ways to sort of like cuz I. Think a lot of folks like I, see people jumping the gun, because they're eager to get this done, which is great and I, feel like we're not ready to really give them the right feedback and vote on it and sort of make it done done. You know, is.

A

That spreadsheet public, the one that maps steering committee members to the SIG's that they're they're assigned to they assign themselves to it.

B

Is not posted in a public place, but I think it is publicly viewable. Okay, I mean we can advertise it, so they know which steering committee needs to go bug to learn about this stuff. I think that would be a really great bi-directional thing: I just don't know if we're ready for that yeah.

G

I thought that was premature good and then we haven't even settled on what constitutes a good charter. Yet they're.

A

E

So the questions we need to answer is what constituted charter. What is the process for proposing a charter and how do we vote on it? Michael? You know how we closed on it. So those you know and then communicating what the state of the chartering process is to the rest of the community. That seems to be the the things that concretely need to be done here. Is that sound right?

E

We had some owners to this by we I mean I'm.

B

Over committed so well, I can go. Take a look at the spreadsheet and a sign I can put a bullet for each of the states that have Turner's to have people next to them and sign for that way. If assigning people in this meeting net stock works for you to track this and then I need somebody to. We need somebody to own the like meta process on like how do we close it timeline and stuff like that.

A

I can I can own them and press that stuff if nobody else wants to Thank.

B

You Michelle, what.

E

I would love to do with some of these things as Michelle took that action item there, and this is my experience from things from here. Can we get somebody like you know a reviewer for Michelle, so that has to get stuff done? That's the person that will say: okay, we've had enough discussion, let's just get this. Otherwise there seems to be a long tail.

E

H

Can I give you the reviewer for Michelle, okay and we definitely wanna.

E

Get our wheat because I'd be okay, make.

H

More progress on this stuff, it's tough to get stuck just just a quick meta, just to get a quick voice vote or whatever. Where is everybody else? Okay, with that proposal there, that joe said basically we'll have one nominee reviewer you get I, don't know a week for comments or something like that close. Let me go I would.

B

Like for us to circle on that, like we close it, if the next meal would be my only thing so I view, you I would view the reviewer as the first passcode. Would you use the authoritative guess now I'm.

H

B

But I don't want.

H

To see a situation where we're forced to bring it into the meeting right, like I, prefer that people looked like, we can say: okay, you have till the next meeting to raise anything, but I'd prefer that we do it synchronously on the review process, I.

E

Mean maybe it's like that's the person that calls last call right. It's like boom last call. Let's just get this thing in that works and I'll call it. It.

A

Wow to be come to a conclusion without having to be timeboxed all.

E

Right we're trying I'm.

A

E

This stuff can get super frustrating right stuff here, there's been, you know, I know that that Tim and Quentin you know, got a chance to meet and talk about this stuff, some more I I missed out on that. Do you guys want to give your current snapshot, because we missed out last meeting with forum, we went back.

C

And forth a couple of different ideas, and we we kind of went back to the original simplest idea- was to just use dev stats and you can get a rolled up. Dev stats across all of the different kubernetes repos for github events and just drawing cut bar line within that. Pretty much gives you a rough guess of people who are active in the community and there will be some potential bad actors within that whole state space, but by and large the signal-to-noise is higher there. So I think.

C

Just by having that without having to do a lot of people process, it gives us a good measure of a group that we can they. You know they have voting rights within the community. It's about on the order of 600 people.

E

So we're separating this from any idea of sig membership, which creates a certain level of cleanliness. There I think part of the proposal. There was also to have a a a nominate. You know: hey I'm, not on the list, but I think I should so that there is room for for somebody to be to nominate themselves and I.

E

Think the other idea was that if we see somebody doing something stupid like submitting like a gazillion PRS with the one letter changes for spelling fixes- and we think that they're gaming- the system- we also have the ability to sort of raise that issue. If, if somebody brings that to our attention and and say now that doesn't that doesn't count.

I

Yeah I think the proposed process for exceptions was, if you not on the list, and you think you should be- you get a one of the six to sponsor you and say we think this person should get to vote with a brief explanation of why what you do on that seat and then we would, you know, fit that here, but should mostly be a rubber-stamp process if a sink thinks you're legit and explains why then you're good right issue, NPR.

G

Comments are counted in the distance yeah.

C

There's a general roll-up of github events, which is everything so.

B

In emojis and everything yeah.

C

Including everything which fill demography.

B

C

B

Actually, anybody whose last name ends and thought is excluded.

I

Yeah we had a quick look over the list and it looks pretty sane like most of the people at the top of the list looked like there should be at the top of the list and there's nobody who looks like they should be in the list or the green nodes that should be in the list that were obviously absent and it seemed pretty reasonable. So like what.

B

What specific events, what specific threshold you don't have to hash that out here, but I'm, just what's ballpark all.

I

Events- threshold 100 issue and.

C

There was no order of six in two people: okay,.

C

Over what time? One year since one year time window.

A

That includes emojis I.

E

Think some of the thinking here also is that, like you know, do this for one election: let's see how things work, let's see, if, if we think that things are distorted and then like there's room to to adjust yeah.

I

The worst case, failure mode is actually really not that bad. We get like three bad actors on the on the steering committee, which is not anywhere near enough to really cause big damage and we go oops and we dramatically overall at next time, but we hoping that that won't happen.

G

Tenpoint, it.

B

Can we see that written up for next time.

I

Yeah I think Tim's done most of that already we can just tidy it up and send it out. Yeah.

C

I'll make a PR for next time. All.

E

Right who's gonna be the last call caller for Tim's PR, okay,.

G

A

G

Terms of last calls my github notification volume is unmanageable. Are we gonna do last calls on the mailing list or something yeah.

E

That sounds like a good idea: okay,.

G

E

You all right, CNCs conformance group. There was the issue here in conversation and Sigma testing shows that they are free running with with do you mean no proper feedback loop? Yes,.

C

Okay, are there words they weren't, so I guess the whole question was we the steering committee had okay because I remember, we called a vote on it. The steering committee had okayed that the CNC F would, you know, hire a bunch of folks, but what was not clear was what was the mandate by which they were given to execute, so that was clear to the community. So what had happened?

C

What I transpired was there was private communication between Dan and some folks at Google, and it was not clear how they're executing, and so what happened is a bunch of PR showed up and they showed up the state testing ax thing about it and the handoff of what the expectation is we're about. How folks were supposed to operate was clearly not articulated.

C

So we had talked about it as a group within the steering committee that they would talk with cigars that bought a test plane and then it once cigar had okayed it. Then they would basically go ahead and write the tests and sig testing would validate make sure that they're saying so. We as a group were okay with it, but there was some drop in handing off from when we had agreed to this to how this the CNC F had appropriated folks and how they communicated expectations and whatnot.

C

So I thought that was kind of odd and it's worth raising to the steering committee.

B

Yeah and I totally second, that I chatted at Dan with length until the fear appeared in his eyes, like I, like I, just using it as a model going forward. Maybe I didn't have the proposal in front of me or something that I think Django or somebody authored that maybe described like what they were going to do, because if so then perhaps I missed the document. That said, okay first, you contractors should create a test plan, and then you should run that test playing through sig architecture to make sure you're, actually testing and conforming things.

B

And then you should go write, some things up and work with folks in the testing common sub project, since they care about making sure tests are often cleanly, and you should talk with the folks. It's like testing about how we're going to run these tests and block on them like that process. Didn't seem to that was all news to them when I started articulating that and say testing, so I would have assumed that somebody would have over that process going forward, and it was unclear to me whether or not like.

B

Okay, we got the money from CN CF and now comes back to the steering committee to make sure it's executed on cleanly or like what the next step should have been. There.

I

It's to be agreed, we need a one person from either this group or sega architecture. Who is the sponsor of this compliance program, given that it is compliance with communities and that person can I mean there is a working group. These people get together. They discuss all these things, they do stuff there. Yeah must were present there and now, after the fact, we're concerned that they might not have done the right thing, so they.

C

Miss back-channeling gate, we'll just did a ton of back-channeling to get things through. It wasn't like to me.

B

It seemed like there was no project manager. That's why I like. I want to use this as a template for like, for example, if we had, we put together a proposal to have automated checking it's a licensing, and that was something that C&C f did like.

B

I would want to make sure that they had somebody project managing that and that that was a plan that we agreed to and agreed with the execution of and I just I, don't know if I missed that plan, when we were asking for the money, I think what because I don't want us to be the bottleneck, that's the bottom line, like I'm cool. As long as somebody gets it and handoff.

B

That's the part that didn't seem to happen is somebody who didn't have the knowledge to put together the plan got it or there was no plan in the first place, yeah.

G

And I'll chase this down and looking at the issue and the PRS can kind of tell who looks.

B

Like in my conversation was Danny, it seemed as though, like he just talked directly to somebody. So that's.

G

B

G

B

His usual style, so I just gotta, make it clear that so I made it clear like please don't ever do that again, that's the exact reason. We have the steering committees to make sure this sort of stuff is done and out in the open. But to me as long as there's a document that reasonably describes what's supposed to happen, we don't really need to be involved as far yeah.

G

So the way other conformance tests have been working is that I have given high-level direction about where we have coverage gaps and you we'll go off and write tests.

G

If entirely new tests, infrastructure is developed, they're supposed to work with SiC testing on that, and then the the actual tests that gets written should go through the sink to make sure that are functionally correct, for whatever the functional area is, and it goes, it has to go to stick architecture in order to actually get into the conformance suite, because the only people in that owners file are Clayton and me yeah.

B

It was just in this sense, so in this specific scenario, they were coming to state testing asking us if the subject matter of the tests was appropriate to test the conformance of kubernetes and that's a discussion I feel like they should be having in cig architecture with you're correct, and they should be having that before they ever write a line of code yeah.

E

And I think there's a difference between sort of encouraging community members to fill in gaps versus you know having this paid relationship with contractors that the CNCs hiring yeah.

G

E

I'll see if I can also as a point of order, I think we should look at sort of why you and Clayton, by the way, because you know I, would expect. Maybe you and Jay since you're. Both sort of you know chairs of the committee, but this is the chair versus technique thing no, but I, just like in terms of conformance like you know, you know, Tim mr. Sinclair has been like super involved in conformers for quite a while. You know, combined with with the work around some.

G

Ways so the rationale originally was that we took a subset of the API approvers yeah.

E

But conformance is okay, different discussion I think we can discuss that in sig architecture but yeah. You.

C

Know I think the point of order here is about like process or lack thereof. We.

G

Can change the list of people but because we didn't have a clear. Let's.

E

Set that aside, that's just me getting off topic. I. Think. The big question is that, huh as as a CNC F hires contractors, how do we make sure that they have good oversight? Yes,.

H

Yeah and I think I think that probably sort of pushing them through the same proposal process for any other kind of feature is probably the right way to go about it right and so sig testing having it like a plan of action practicing testing. This probably would have been the right thing here. Yeah.

B

I think somebody within I think these were ended up, giving it as a p.m. with thinking goal, but to go back to the higher like point of orders process here. Like did anybody else? Read the proposal for wire easing as funny did it? Did it elaborate.

H

It was this. The coverage was this: just the coverage like API coverage testing proposal, or is it something yeah so I had talked about? This was more accredited. Okay, cuz I had talked to Dan about that, but it was that. Could you elaborate on what you saw there? We were talking about basically building a proxy that would sit between the API server and the test infrastructure that we had and would measure effectively coverage on the swagger spec right.

B

Yeah, so that's something I've seen I've seen demos of something like that from folks at stake, testing and I think they're funded by CN, CF I'm, not sure. But again it was unclear to me, like who's sitting here, saying like in order to most effectively prioritize what tests or API is what does need to be written. What api's need to be covered? We need an API proxy thing and then, based on that we're gonna spend these like I just didn't see anybody. You.

H

Know with weed so.

B

Having that so Ida Isis.

H

Specifically I had had those conversations to a Dan at one point where I said: hey I've wanted to build this thing for forever and and yeah. You know, I think that was the impetus of it. I, don't think it was ever sort of I didn't, say: hey. You should go hire somebody to do this. It was more like he was looking for projects to hire. People for and but I agree that there's no process right, like I, mean I think in general.

H

This is true for the money right like there's no process for prioritizing what we spend.

C

H

D

E

One of the places where this fell down, in my mind, is that I talked to that person and I proposed the alternative instead of a proxy as I was talking to. It occurred to me that we can use the audit log as a way to actually look to see, what's being called, which I think someone else.

G

D

E

Meta brennan is that we didn't have clear goals that so we couldn't actually figure out whether the audit log or the proxy was the right solution for the problem we were trying to do because we didn't have a proposal process in place here. I would.

B

Like to hear from Michelle who said she had, are you.

A

You know I want to let this continue. Okay,.

C

A

When I come back to the beginning and kind of from a high level, take it from the inception to what happened and kind of, not that again. I.

E

Think I think we're we beat the horse here, everybody's had their say. So why don't you go ahead? Michelle.

A

Yeah so I, so I read a proposal. I think I talked to Jay go and I wrote a proposal. Aaron you were in that conversation we talked through how it would all work you're like yeah, that's something, and then we brought it today. I talked to you this steering committee about it. Brian grant gave us some context. We're like that sounds good. We like got this money or we got approval to get this money and then I had like no idea that people were hired.

A

So when people were hired Brian, were you told, like what stake architecture told.

G

Typically, but before you got funding, they already had that contractors identified. That's true, so I didn't know that they were starting right. I knew the people had already been identified. That check. That was one of the arguments for the urgency is that you know we don't want to wait for the those people might get hired to do something else, and then they won't be available and then we'll have to wait even harder.

G

But yeah I was I was not aware what they had started working on her when they had started working so.

B

I do remember green, especially with you and Django, and he'd mentioned that there was a proposal, but I hadn't seen it, and it was unclear to me whether, like as a process going forward, if a proposal of that nature should be floated through the steering committee or if that's making as a bottleneck, I.

A

Was under the impression that that proposal was floated through sake architecture, but in the future, yeah I think that it doesn't hurt to pass that through the steering committee. There.

G

May have been a link shared with the steering committee, but I'd, probably you know there was there was discussion about it, but it was high level, but the proposal was also high level proposal was along the lines of we have evidence. We have data that shows there massive coverage gaps.

G

We think we can contract people to fill some of those gaps, but they asked for a rule that the project would require people developing new api's to actually implement appropriate conformance tests, so he's not in a hole and you did not be digging a deeper hole at least and then, like money was allocated with the.

G

Expectation that the details about specific coverage, gaps to be filled would be worked out as they developed tools to get better information about what the actual gaps were like the proxy or whatever so I had suggested that that be the first order of business. I was aware that data was collected using the audit logs, which I don't know if anything's happened with respect to the proxy.

G

B

That's all I'm going, but I don't want us to dive back down there. I really like staying up up here, so I feel like what I what I wanted out of this before I would want to send money through or at least as a confirmation, once we had an approval to move forward on a project.

B

I'd want to see something that laid out like who was owning it as manager, whatever person shepherding, experient and I would want to see which SIG's were involved, because I would hope as much as possible that this work would be done out in the open and through the appropriate six that have ownership or authority in those areas. I just don't feel like I got that with this particular case. My.

G

Impression that was that it was supposed to be. You know, as far as like any new coverage measurement or test and prescription that that was supposed to be discussed in the cig testing. It.

B

Like the implementation of it short but right, the the whole idea of like you want to make sure the tests for the work load stuff, her run through, say gas, because those are the people are the domain experts in that specific area and like I I, think that's really. What I'm interested in is a steering committee member is not the specific technical details, but just making sure that we're routing the right work to the right sinks, yeah.

C

It's basically like if we're gonna hire people make sure they have well-defined tasks and how to execute. Otherwise, it's just kind of random chaos right like and that's kind of what we're. Having is random chaos, so I think. If we're going to spend this kind of money in we should have, we should have the due diligence or have insight into enforcing the due diligence of some sort of process right because it seems like we just said.

C

Like you know, money, random things happened, and then all this and a bunch of questions right, that's the way it that's what happened afterwards. So there was, there was clear lead up, but lead up was fairly clean. But then, once once the hiring of these folks have occurred, then it was I, don't know what happened in.

A

The process just to go back to the process of like leading up to getting the money so that requests for money should that be talked about in the relevant SIG's as well like before. We actually get the money. She, the requests, be talked about with the with the people involved, with their groups involved, or should it be talked about after the money is allocated or.

C

Both I think as long as there's a plan right some kind of plan, as part of as part of the you know. Having if we're gonna say we're gonna hire these people. We need to have some pipa plan that when we hire them they can, you know we're contracting them. For this reason here you go. This is what this is being executed, not.

B

Job yeah I'm, not I, wouldn't want to be like voluntold in a sink like hey. You must now go work on this. It's more like hey guess what you're you know. Have four people dedicated full-time to working on this task? Do you mind making sure they're doing the right thing.

A

That make sense.

E

Are there any action items out of this, or is this just buyer? Beware like board to the wise for the next time, I'm.

G

Gonna chase down with so I think Mithra has been.

G

Interacting with the contractors to some degree and had found a doc I'm, not exactly sure to which groups I thought, it was stick testing.

G

With some high level, I wouldn't say details but sort of a high level about what they're doing and I see. She commented on the issue that links to the pr's that are underway, so I'll try to get a from Mithra and convey some of the things we talked about here. There's a process.

E

Do we want to require some sort of check pointing where we actually evaluate how these contractors are doing over time against a plan? Do we I mean how formal of a process do we want to actually get here? Well,.

G

I think, if they're going totally in the wrong direction, we can reset them if they seem to be going in the right direction, then putting some kind of guidance and process in place with where we make sure that this stuff is getting routed to the right six yeah. One of the first issues they created was a test for namespaces. It's probably not even obvious for them, which cig that is and definitely part of the challenge.

G

G

Let me figure out, what's going on a little bit and then we can figure out what I'm.

E

Just trying to like can we extract some lessons out of this because, like if you were at a big company or a company at all- and you were gonna- hire a contractor to do something, there would be an owner on the company side and they would be the one on the line to actually make sure that that contractor was was useful and they would actually I. Think.

G

There is an owner, just the owner is not maybe doesn't fully understand the expectations. Yeah.

B

Whatever, like Joe I, do kind of longer-term what that more formal process or document or checklist, we would go through going forward. I. Just don't think we have any other proposals in front of us right now. So I feel like we're all on the same page and we reach the buyer. Beware state and.

G

B

And you'd be a good thing to do, but I think we're all kind of yeah.

G

Just as contrast so the previous things that we allocated money for the dev stats effort, they worked pretty closely with me and with Garrett. We basically got daily updates in the dashboard and it was developed. There was an initial dock where we flushed out some level of detail data. We thought we wanted and then you know we found when we actually saw graphs that the data was bogus or that it wasn't actually he's full or whatever, and then we iteratively refined from there. But there was a closed feedback loop.

G

The people who actually knew some things about.

F

G

Was actually needed and the other case was the editor for the docs and they got embedded into sick Doc's, and that seemed to work pretty well, but they're also, you know like they actually knew what they needed and there was a tight feedback loop. So that seems to be what's missing. Here is both actually those myths, I mean.

E

Countability up through this group would be would be useful at some point. All right. You.

H

Know, I really don't want to be the business of p.m. in this right, like I, think that we should I think the right accountability is to say we're not going to fund anything without a cig sponsor. Well.

E

But I mean then the big end up p.m. unit right I mean like that's.

H

I mean that's okay, right because the theory is doing benefit to the cig and so like. If you, if the cig it doesn't want to PM it, then that's a pretty good indicator that it's probably not worth funding well,.

G

So, like I said in this case, I think there is someone who is involved in this thing, who is actually doing it, but that person just doesn't fully know what was expected all right.

E

Okay time box that we've already went over because I lost track of the clock, Michelle go to conduct committee doc. You have four minutes I'm. So sorry, no.

A

Problem, hey I, just wanted to make an announcement. There is a code of conduct committee outline for how to move forward that why what the objectives would be and how that would go about and yeah several people have started commenting already. So I really appreciate that if you haven't seen it check it out, that's it there's.

G

Also, a working doc to improve the actual code of conduct that was started on the CNC up side and I'm kind of driving it a number he who expressed interest in participating, the doc has been shared with you. I could share with other people, but you know really, if you expect to work on it, I will share with you. Otherwise, it's not really fully public. Yet at this point, I just like collecting the issues and trying to figure out what the shape of what is needed. Alright.

E

And then the idea is that we'll keep that private for a little bit yet well, we, you know, have a safe space to iterate on this stuff before we go more public in terms of what the ship of the committee looks like. Is that the idea all right, yeah, all right, Lee? Three minutes? We want to continue arguing about the conformance stuff.

I

B

Got one last thing: real, quick with the time remaining. Just to give you all a heads up that I am dropping off the face of the kubernetes hearth for a couple weeks upcoming now that I've assigned you all actually is my my last day at Samsung is Friday and I'm. Taking a couple weeks to recoup and recharge I will be back, but if there's anything that needs ándale urgently, please get in contact with me over the next two days.

B

Otherwise, I'll be checking email and other forms of communication sporadically, but I'm gonna try as much as I can not to be around just figured I wanted to tell you all to your virtual faces since I've enjoyed working with you so much and I'm going to miss you all for a little while, as I will be back soon. These.

D

Two weeks just go I think you'll be fine.

I

Five minutes at the end of the meeting, just two I think we've been talking past each other a little bit, so we can fix that in two minutes. If you have two minutes, oh.

F

No I have to run brian has to run he's our he's. My zoom link and I can't get zoomed to work on my laptop right. Now we can switch to hangouts or something yeah is that.

C

For the other thing, because I added to the agenda for cigar no.

I

It's not anything to do with cig architecture or sake anything its minton I'll.

F

See you guys later and I'll, let.

E

You in Jim underway.

F

Did you wanna, hang out to me real, quick sure I will send you an invite how about that Roger cool thanks! Well,.

E