Kubernetes Steering Committee, 8 Nov 2017

Previous Meeting

⏯

youtube image

►

From YouTube: Steering Committee Meeting 20171108

Description

Recording of the Steering Committee meeting. See also:

https://github.com/kubernetes/steering

A

Okay, so we're now recording it's 107 on Wednesday, November 8th. This is the steering committee meeting and we have nine folks joining I. Think you can, you can see all of them all the names. Let's see cig testing is Erin and sig note is, is Quinn. No, who said note network.

B

A

Is Eric I'm, sorry, I'm, Eric yeah, no worries I. Just we need the decoder ring. Whatever we have, we show these things yeah so go ahead. Take it away art, okay,.

C

Cool, so just some points of administrivia around we've, all, probably speaking what I'd be as public as possible. I want to make sure that we make that our meeting notes can be public. The question is whether or not the existing meaning, not notes stock, that we have right now. Do we just go ahead and make that public, or do we prefer the idea of copying it to another doctor posting to the repo?

C

The only argument I would have against making this doc public right now is as we're taking notes which, by the way, can I get somebody to volunteers, a note-taker, Joe and Brandon.

D

Have volunteered.

C

Nicely for this in the past, so maybe somebody else thank you Sarah. um It could encourage somebody who, like somebody to just like watch the meeting, live by way of notes which, whatever, if they're, really not interested I, guess I can or any any disagreement to just making the existing meeting at a stock public to this I guess: kubernetes down read-only.

A

I mean: can somebody volunteer just to do a really sort of quick read through and make sure that nothing's gonna.

A

The biggest concern that I have would just be you know things being misinterpreted, you know without without full context, yep. Thank you. Okay,.

C

um So next thing would be since we're recording this meeting we're gonna want to post it somewhere the way I that I've historically done it as a sake, is I just post the meeting to my personal account and then share it so kubernetes, playlist Joe, you happen to be recording today. We could just say that whoever's recorded meeting posts to their personal playlists. On the other hand, I think the kubernetes community meetings are posted through some, like special kubernetes account. Do we care there.

D

Is a kubernetes YouTube channel which we can host it directly at, but it is no different than are not substantially different than creating a playlist there and having us post them wherever they post and then using the playlist. Almost of course, you ever want to clean up your your YouTube.

C

D

And then I think that's the simplest method. I mean.

E

Does this like, if I, if you know, is there I hit by a bus issue here, I mean not that, like in 10 years, anybody's gonna want to watch these movies anyway, but like it's.

A

All I'll talk to George I know he has credits to the YouTube channel and I'll get this stuff posted. Ok,.

E

That makes sense so.

A

You can add an action item for me on that. Ok,.

C

Yeah, ok, I, don't really mind that people who watch state testing meetings get suggestions to watch videos of me drumming so.

C

Ok, there was an issue filed against our repo from the CN CF asking. If we've nominated somebody we volunteered Michelle was tribute, do we need to formally hold like a public vote on our mailing list or anything it was. It was announced already, it's kind of a pietà copy, responded on that issue and said yeah, it's her when you say announce, wasn't any other well.

F

It was announced on seats, multiple scenes. Yes, mailing lists, ok, great cool and Twitter, which is you know.

C

That's official then yeah, that's put a bird on it. Okay, so Brandon volunteer to be the next person, I'm happy to be posted for one more go or handoff to Brandon. Does anybody care in particular? What did we decide? The granularity was. He was basically every month, so we.

G

Started mid month, I'm wondering if we should just continue until the end of this month and then start that's well. I'm.

C

Happy to close out another.

F

So no meetings in December are there. There aren't well I'm, not gonna, attend to me because there's cute con yeah, those Keep Calm.

A

There's one scheduled during coop cute con I assume that we're not gonna do that and then there's the 20th which I'm gonna be out. I, don't know, I mean we should see. If we should take a we could see if we can make quorum there, I can make that meeting. I'm not planning on being out there, but like I, don't know how other folks are doing handling it. Yeah, okay,.

C

Let's not decide that today, but I will close out November. Okay, actually, I didn't follow up, so Tim literally created the mailing lists. We were talking about last meeting as we're talking about them, which was super awesome. I feel like we might be vaguely unclear on whether or not our document to the policy for one public and one private I saw some discussion on the public-private mailing list from Clayton, but I'm like hey. We can. Maybe we want to make it clear to people that they could post to steering private and then we could.

C

We could advocate that hey. Could you please take this public? It seems like it's it's worth taking public a do. People think that's a good idea. B. Does this mean we need like a poem request to contributing or something the contributing and D file that Jo so nicely put a poem requesting before after last meeting.

B

Did we have a forget? Did we have real use cases for people who'd want to contact us privately, that wasn't co-ceo or security I.

G

Think the other I mean people would end up like contacting us individually in.

B

The past yeah yeah, but like is there any topic that that we endorse them contacting us privately. For that we wouldn't say this should go to the public list. The.

C

Example I gave was where you wanted to ask in about like the disagreement between two SIG's or two people within two SIG's: that's not really code of conduct, because it's a disagreement and you may not want to publicly go and say: hey I've had this big disagreement between 60i, you might very well but like I, get that a lot which is hey. Can you negotiate between these two points of view and I would usually say?

C

Like you know, if it's, if it's personal, let's get a broader group involved, it's not quite code of conduct because that I think they're gonna be people who might see code of conduct is only people who are actively seeking to do harm versus disagreement right.

A

I'm, okay, having it you know and and saying like hey, you know we're all possible for you, yeah I, don't think many people use it and then, if it's not useful.

C

Like it doesn't hurt anything well.

F

I think that's where I also get a fair amount of that, hopefully having a formal escalation process will at least for technical issues I mean, maybe not all disagreements are technical for sure, but hopefully that will address some number of these. At some point, though yeah for now, maybe just telling people they can send to the private list if they have any concerns, is a fine thing to do.

B

Okay objections I'll make this change to make it post a ball.

C

Go for it can we get somebody to take an action item to pull requests to contributing about our deep right now, so Stern does email the steering and it sounds like. We just need a sentence that says if you need a private discussion. Emails to your in private words aren't coming to me right now. I can.

G

Do that Thank You.

A

Michelle feel free to just slam. It in I think we're all no need to go through formal talks on that. Please pull request full request on oh okay. Well, you can merge your own pull request. Can we say that I'll.

C

Be around that just forever standing! It's fine yeah.

E

But I think merging I think merging your own pull request is fine I, like the audit trail of having pull requests, but I could sell for just one.

B

So, on the same topic, there were some on unclarity as to whether the steering public list is subscribable by people or not.

G

Energy injection I, don't.

E

See any reason not unless we somehow want to use it for an off-grid somewhere. So.

B

My my only fear- and it's currently implemented as subscribable and a few people have subscribed, and there was at least one thread that was involving matt. Who is not a steering committee person right.

B

My only fear is that it just turns into committees dev and that we lose focus as to which ones for what, like honestly I thought, the colonel's distributions conversation could have just easy just as easily gone Archimedes dev is.

A

Second, that time, that's kind of my concern is that I, you know the more people on a list, the more it turns into sort of rapid-fire boom boom boom type of discussions, and that's that's just gonna- create gonna, be harder for us to focus on stuff that.

E

Sounds reasonable to me: I! Don't have a yeah that.

F

Seems like a real.

E

Risk so that's fine.

B

Sorry, so is there a decision? I, don't know what is changing the policy front. Yeah kick people.

E

Off learnt on the committee and closed it I I.

G

Don't mind people being on the thread and I think if it's something that people want, the steering committee is opinion on I. Think I mean it's a valid question, but I also don't feel super strongly about it. That's just money, respect to so everybody else wants to close that. That's find you.

A

The other thing that we can do but are going go ahead. Circle.

H

I mean we're getting back to this, can you read or can you write issue so so clearly we want people to be able to communicate to the steering committee right. So that's private is is writable by everybody, so anybody can send to the steering committee and nobody can read that and then there is you know: what can people can people get into discussions and respond to discussions on a given topic and I? Think that's the problem we're running in right.

H

So isn't that solvable, but just not having I. Guess you don't yeah there isn't a permutation to breeding right that solves that problem other than shutting down the list. Yeah.

E

It's not post, you can post, but you can't subscribe, which means you're, not gonna, get random responses, but you'll see your own posts. You can also yeah.

F

We could make all of us managers of the list and we can say, managers can post alright.

A

So here's another idea: how about we say you know the benefit that mean. The purpose of this is for the steering committee to benefit. If discussions are getting out of control, then we'll, like you, know, retarget them to the appropriate list. We put that in the contributing also and then, if somebody's really sort of like being super randomizing, we just sort of have a private conversation. Saying hey like we really need to keep this list. For instance, I would prefer that approach as well. It.

G

C

Agree, the some of the discussions got pretty passionate pretty quickly, but I would have seen anything wrong with somebody saying it seems like it's out of scope for this list and just like these sorts of examples will better help us to find what the scope of the list really is, because we don't know we're iterating, so yeah. So.

A

We leave it as a document that, in the contributing also I am.

C

Okay, leaving it as is and yeah something contributing saying, we have.

D

C

To suggest that the conversation is causing too much noise for what we hope is a high signal list.

C

Okay, anything else on the mailing lists and how we use them all right. Next up the Charter, we all had an actually the last time to go, read the Charter and decide if we were ready to ratify it. um The gist that I took away was everybody is like yeah, I kind of makes sense, but also a lot of it was really bootstrap centric and we like some way to create an amendment process. I, don't um I.

C

Think I saw a suggestion from Tim Sinclair about just straight-up, make a new straw man to try and remove all the bootstrap stuff. I saw something from Quentin about. Couldn't he just make a simple amendment process and I saw something from Brian about hey what about? If we use my modified lace consensus thing, what can we do to drive this to completion and actually ratify the Charter before kou : I.

I

Think just stripping down the main one for its basics, because I think in my email, I, don't recall but I. The Charter should have like three main topic areas: I think I wrote them down. If not I actually got some homework. I forgot to bring notes, but we kind of went beyond that in the scope of the original document, and it was meant to be as how do we bootstrap this individual process?

I

If we rip that out and just keep it yeah mission scope and outline if we keep that very concise I think we can, we just move forwards and maybe pumped off some of the other election things to its own doc. Right.

G

Yeah I'm in favor of doing the election stuff and that's undock I agree.

B

A

That I think the biggest the biggest thing that I think we really as a group need to nail down before the the next election is move past sort of our less than ideal member of standing mechanism into something that that is sort more sustainable, but doesn't need to do right now. Yeah.

F

I think the doc is big about.

A

F

Members of standing our chosen right, but.

C

What only concern with redefining the Charter and now is I, don't want to see us go through another multi week, long process of coming to consensus. If it's just this group's responsibility to come to consensus on that doc, that's that's possible, but I would really I would really like to see us have this done in two weeks, because yeah.

F

I mean so right now, I think amendment process wise if someone proposes a PR and it gets algae teams by all of the steering committee members.

F

That seems like a fine starting point and then, if we need to relax that to make faster changes in the future, we can refined, won't come up with a voting scheme or something.

C

So if we were to do a PR, would that allow other members of the king of the kubernetes organization to comments on that PR.

D

Would be able to write, feel.

G

Like they wouldn't really be interested in that yeah.

F

You're one people to comment orders: you don't want people to problem.

C

I eat would rather I I felt like there was some pretty long comments trans on it last time, because it's gala on longer than it need to I mean I, don't even great work clear that it's really this groups responsibility and it come to consensus on it, not yeah.

F

I think easiest thing to do about that is just don't advertise that we're doing it. We actively solicited comments multiple times, yeah on the original document.

C

There might be some who would say those comments, weren't address. We just moved forward anyway, which is we're.

B

All considered they weren't all addressed.

F

And that's where we have to understand: okay, that's.

C

For a pull request and LG teams from everybody like the most same thing, I just wanted to point out. Okay,.

D

C

Tim st. Clair does that sound like a PR you'd be willing to put up I.

I

Can put up the PR soonish, some of the modifications I was looking at, but I think I think it requires brought potentially a couple pr's to clean. We all agree upon because right now it's kind of bloated and I don't know if that's the structure that we want right, ideally going forwards, so I think we should just kind of be a little more fluid and then at some point in time revisit for full ratification.

I

Because if we try to make every nobody else, I'll agree upon every PR it may not. They may take too long to get done according.

C

To our existing charter, we just need quorum, which would be 2/3 right for.

I

C

Well, what are we using the signify ratification here? That.

I

The end result is complete, like I'd rather have a couple of people who can fast turn on shuffling the doc around to make it look the way we think it should and then coming back for consensus for an interview. So.

F

Don't do that in place? Do I suggest.

A

Do to create a draft file operate on that boom boom boom and then the the PRK essentially swapped in the draft would be the the one that we would call a vote on.

C

C

Sorry I mean perfectly if I'm wrong, I, guess I'm hearing it would be desirable to just throw together a Google Doc and share it with this with the this group, allow private to comment, allow public to view only, and we drive this in the public's view.

B

But focuses specifically on streamlining the stuff: that's there to remove the initial election material, to move that to a set they're docked and not to rewrite the thing right. And yes, what.

C

I heard was we want to remove the bootstrap related craft, we want to defer, redefining and articulating election related stuff, including members of standing, and we want to make sure we have in place an amendment process.

H

Yeah I'm a little worried, I mean we've been we've been at this for a month now and we still haven't ratified the thing and that's why I'm just like we're not totally in agreement what should be in and out I just kind of thinking that all we need is to say we had that document. We agreed upon it, there's nothing in there, that's wrong. We can shuffle stuff around move parts, two different documents whatever, but we don't need to do that to ratify what we already have only thing that I saw.

H

That was, that was really problematic. Is there's no defined process for changing so so that we see if we're going to defer any changes, we have to have like some agreement on what that is going to look like this.

I

The mission statement of the first thing is totally incorrect, though the mission statement and the beginner I'm not gonna hate to like needle this, but it's it's not what this group is intended to do. The the initial statement in the original document was how to formulate the bootstrap committee. That's not what our mission is right. So that's the reason why it's like it's odd, even though the document took a long time, it was for the initial group. So that's not what our arch our Charter should be.

I

Our Charter is not to talk about the bootstrap committee and to get this process started. I.

H

Can have another look at the daka I didn't remember that being the Charter, the Charter said what the Charter of the actual steering committee was but I liked again reread it.

C

But I I kind of side with Quentin in that I think the most critical piece that is not is not. There is a document to process for how we change it that we've all agreed to we literally just like. Should we pull request it? Should we Google docket? If we get that, then we can work on the rest of the stuff.

D

So, who is taking an action item to draft an amendment process and then review from existing language.

I

I'm not going to do the amendment process, but I'll streamline the main doc.

J

We have any strong opinions right now, an amendment process other than like two-thirds or I mean. Is it really that complicated that we need to trap too much around it?.

D

J

B

I think simpler is better. Less is more right if this is longer than a paragraph or two we're doing it rolling. Yes,.

C

J

Be fair to say that, like the amendment processes, an amendment is posted to steering one week in advance and is ratified with two-thirds vote.

F

Also it's some of this discussion. We had earlier on about consensus, lazy, consensus, voting, etcetera with cases like well. What, if someone calls a vote and just and I've heard literally a third of the steering committee is out. This is still okay to have a boat, then I, we could just say well, we'll just trust people for now and start with two-thirds. We shouldn't be changing the document that much I'm fine with that hold.

H

On it's two-thirds of the steering committee, not two-thirds of the people who happen to be present, I mean if you have any.

F

H

So if one-third of the steering committee are not contacted well, then you just can't get a change of proof time.

A

Okay say that will do lazy consensus for like day-to-day business type of stuff, but for like something like amending the Charter I. Think just you know saying we need a straight up or down vote seems reasonable. I also.

E

Think I mean I, don't dude. Do we genuinely believe that people are gonna, try and pull things like that and not get it called like I hope we don't ever degrade to a point where most people would say. That's just too big a decision for us to make with that number of people missing. Even if yeah.

F

Yeah so I'm fine with just starting with 2/3 and some reasonable amount of time.

C

Suggest the FN amendment process I mean.

J

I can type up whatever we just set here, which was one week advance to third one week: advance. No, that's 2/3 vote ratify we'll be done so same. If you want to do that in the same dock that you were going into getting the bootstrapping language out, I'd be happy to do.

F

We want to just say we vote on the public mailing list. Do it.

F

J

Yeah, that's fine too. Yes, both.

F

Ways archived and it's immutable that was one of the issues with using github be ours, is that anybody with write access can change the record theoretically.

C

C

Alright, so I feel like by the next time we meet we'll, have we'll definitely have an amendment process and will will at least have poll requests that we may have come to consensus to you for simplifying this charter, and that's good at the you know feels like a pretty reasonable thing, not okay, okay, unless anybody has anything else, I want to spend the rest of the time talking about incue, bation I'm, not really sure how to facilitate or lead this discussion. I feel like every time we talk about it.

C

We end up trying to solve multiple problems. All at once, I think that I.

C

Would like to come away from today with like some actual things we can do to push the ball forward. It sounds like there are a couple like there are a couple. One thing seems to be a review of, what's in kubernetes incubator and like what needs to be in and what needs to be out, it seems like there's a definition of what the kubernetes project and its architectural vision are at present.

C

I, don't know where that's written down, and it sounds like we're suggesting that the architecture owns that and so you'd have to just go talk to save architecture. There's.

D

C

Problem of endorsement or what is in what is out and yeah the shelter Brian posted that this there's the potential for an ecosystem working group, but as much as it pains me to suggest yet another working group or sake for this project. It doesn't seem like there is an existing body of people whose sole purpose is to facilitate the promotion and discovery of projects within the kubernetes ecosystem and rattled off a bunch of things they all kind of seem to revolve around incubator.

C

How would we like to proceed and breaking the problem down and assigning people to take on pieces of this problem? So.

E

So I sort of see three categories of this. One is what I would call sig sponsored projects which I I think every SIG's you just have their own scratch space, and this is things like you know like the kubernetes autoscaler or the work on snapshotting file systems or there's a bunch of these sort of like.

E

We think this might eventually be part of core, but the sig is gonna, explore it out of core for a while and maybe it'll stay out of and I think those should be mostly freeform and and and and if sig wants to work in one of those spaces, they should go to work in one of those spaces and then I think there's the like.

E

Everybody wants to be an incubator for the official communities part or for the CN CF CLA or whatever, and I think that we should just get rid of all of that and then I think there's a notion of like well. What have? What are about the things that I think that started on the outside like helm, but we I don't want to maybe bring in and I think we could actually just punt that and say hey we'll deal with that on a case-by-case basis, win and if it turns out to be a thing.

E

A

But there's another category, which is projects that are not an incubation now are graduated yet still don't necessarily fit our definition of what we think should be yeah.

E

I feel, like that's a hornet's nest and we should just be like well sorry grandfathered in like it's just is what it is. Maybe we can clean it up, but we got better things to do than worry about that. Ooh.

C

Burn Eddie sword: I, really, don't think that number makes sense, I think.

I

I think we should have a sunset period, honestly, I, don't I, don't agree that we should grandfather things in and just let it slide, because it sets a weird precedent where these other projects then have this weird pull against ones that want to get into the repo and I think if we provide a sunset period with a graceful way to on-ramp into some other process.

I

That gives a long tail for folks to transition, but in a safe, meaningful way and part of the conversation we had last time which Phil wrote a ton of notes on which was really good was the idea of potentially having a separate, org or set of gorgs that could live in it. That doesn't necessarily mean it has to be underneath the kubernetes umbrella, I.

F

A

F

Didn't I didn't hear what Aaron said before Tim started talking oh I,.

C

Was just saying we have about 61 repos in the kubernetes org and that number doesn't make sense to me going forward. Oh I.

F

Actually think it doesn't make sense, then, that it's not or not.

J

My primary concern is that the existing set of repos- we have don't have always a clear chartered boundary, and so, if we let certain things, stay, grandfathered I think it's just moving a conversation from one area to a rather loosely worded.

E

I'm not opposed to um grandfathering things or sun setting them, as Tim suggested I. Just think that that's the biggest fight right, and so if we want to make progress, I suggest that we. That is not the first thing that we should deal with, because I just don't think we'll make progress. I think we'll just get mired. All right aren't.

F

We just discussing the exact same stuff that we discussed last time.

K

The first category that we brought up was things that are owned by a sig the process whereby a sig decides which repos get there is if a sig decides to adopt something. What constitutes the stick, as you know, do we feel that everything today would would be reasonably scoped as happy to do that? Because does it require participation? Is it require membership? Does it require a vote?

K

Some of these people I, don't think we have to sort out now, but I think the hidden assumption, which is that the SIG's are today structured to you know, make a decision about what can or can't be in the repo, but just as an example like the node, if sig node says one tenth, a new container runtime into Connecticut.

K

We have, you know a fairly concise sig, but some of the looser SIG's might not care or might not be swayed by that. So.

E

To be clear, what I was suggesting is that we have kubernetes sig node as the top-level org, not kubernetes, kubernetes or kubernetes singular.

K

If Caray is node adopted, you know hyper-v, just as a non contentious example. Is that something that sig note can do and I think I.

F

Don't think that's code that doesn't belong to us so that that if it requires transfer of copyright and CRA analysis and stuff, like that, that's a totally different problem, starting a new project with new code. Okay.

K

I was inserting a new container. Runtime is totally within scope for signo and I would agree if you could convince enough people to go, do it and that that I think bounds it a little bit more right.

E

And that was what I had in mind was like. These are the projects where the sig itself has decided. It's something it wants to go. Do not they're gonna, go adopt something that's out there, but it's like no. No, we think we want to go. Do like I mean sod is doing this with snapshot files right, it's an exploration. They don't want to do it.

K

E

Court they have a separate project work. Maybe.

K

Another a contentious example: that's very useful in some of the the apt def working group, the plugins that were kind of talking about. As prototypes we'd say. You know it's very reasonable for six CLI to go, create a number of prototype repos, underneath six CLI, and just because it's under the CLI sig doesn't mean that you have to use it. What's the boundary there right.

F

Right now, one issue we have is there's no clear notion of deciders for each sig.

J

And my issue is that yeah I really mm-hmm, you know I feel like it should be tied somewhat loosely to owners in Cube, but unless for how others feel in the no.

A

I think it could be I, you know in one way or the other. Those things should be related in some way. I I think that's not the discussion that we're having today unfortunate.

E

I think I mean I honestly, like I think that the the goal of making every sig have its own github organization is actually to decentralize it so that that's so that like break the tie, so that it doesn't feel so that the official miss comes from the sig not from the broader kubernetes community. Well,.

A

And we can also, we can also promote things out of a sig into the main Cooper name. All.

E

Right exactly exactly I think that's totally fine, and you know we may very well do that kind of thing, but I think I'm.

K

Good, that's very reasonable, I think and and something under six er kubernetes sig CLI is a lot less contentious than you know: kubernetes config management or kubernetes container engine, or something like that. Right.

C

Just as a strong man, it seems like kubernetes incubator today is that playground or that that safe organization, that collectively all sakes could have dumped any repo that they want to, which is in fact what some of them do now hang on. Let's forget the word incubator for a second like because I'm trying to think about what what you proposed without having the overhead of managing 26 different. You know.

A

But I think with incubator is that there is a graduation process and that graduation is a right not is done sort of as I'm I'm just proposing.

C

Like what, if it was just one organization, forget the concept of an incubator process, just one organization into which SIG's can no.

E

I, don't I, don't think so, because I think that the trouble is it's either kubernetes contribute um stir fire or it's or it's there's this expectation that it's the official thing right, like it's, the incubation of the config manager to become the official thing, I think by giving each cig their own repo, you break the association with the broader kubernetes project and the sort of like I want to be the official kubernetes version of X issue. So.

C

I, just you also really increase the management over at the project as a whole. Why.

E

This is all robots right.

C

Right with the robots, no, but the.

K

C

Bought also a non hang on: it's not just robots, it's people and process, no.

E

Like I mean like looking like kubernetes client right like yes, the process, that's on communities. Client is not nearly as good as the process. That's on kubernetes, but it's good enough for the use and it has the CLA BOTS and like I, don't think setting up could all the kubernetes client repos has been particularly onerous on anybody. You know.

K

It devolves a little bit of the. We do want sake to be a little bit more self-directed nice. It can't be this hierarchy of centralized. You must go. Do this big testing potentially should be able to offer feedback that if we go do this in too much of a direction, we're gonna cause ourselves lots of problems, but I do think. We have examples where a loose model works. Okay, early for prototyping, so.

J

What is the level of test infrastructure services that a cigs organization sponsored project could consume from the cube organization? Okay,.

C

That is an interesting question, but I feel like we need to drive to like I.

J

Feel, like that's like concretely tied to like what is happening all right, yeah.

C

Incubator there.

J

Is inconsistent.

C

Whether or not joining like a kubernetes organization gets, you gets you like some consistent, tooling and process and blessing from kubernetes and stuff is all like orthogonal to the incubation process. It's it's not really a goal of trying to solve bright.

C

Well, what would I would I took away from from our discussion about incubation last time was that we really need to unblock people who are contributing code, that is in the scope of the kubernetes project and its architectural vision, who today are stuffing it into kubernetes kubernetes, and we would rather find a fast track to ensure that they are being good citizens and starting those projects outside of the main kubernetes Rico.

C

But we don't want them, starting in humanities, incubator, because everybody seems to think you ater means not ready yet.

E

C

Ask like people.

E

Don't think they can just start it in a new org and I would say that the reason they don't want to start in any new org is for some reason they want the official blessing I. Think that's actually something we want to break so.

C

We really do want to break the official blessing. What.

E

Unless it's a sig that has already decided like we are gonna, go, do this officially as a sig and that's why I think that the right answer is to say we have sig orgs. Those are the things the sig is proactively decided it wants to build, and then we have everything else and that's the things that people have decided they want to build and and and there is no other thing right and then there isn't a sort of like this quasi official, but not official sort of situation. I'm.

J

E

Brennan is there a kubernetes, org and then there's kubernetes SIG's orgs will keep the existing kubernetes org right that we have that's fine, but then every sig you'll have kubernetes sig, CLI or kubernetes sig API machinery or sig sig, like it's nice. If.

A

We break out cute control. We would break it out in a queue: Brunetti, slash, cube controller, Kuban, airy, kubernetes, six CLI /q control. Well,.

E

I think there are some things that are already part of the kubernetes repo that are already official like that that you might want to break out as cube root n, slash, cube controller.

D

So I'm thinking.

E

About it's more, like the projects like I, mean I, think as I said, like I, think the snapshot, the file snapshot project is a really good example. Where it's like it's an exploration. It may become part of the official API, but really it's just sig storage, that's playing around with it and they need a playground to play around. With that thing.

C

Here's what I feel is lacking from this discussion right now, but I would like thanh us to have by next discussion. I want to understand what the architectural vision is for Humanity's that helps us define what the kubernetes project is with a strummin of like what repos scattered amongst the kubernetes org and the kubernetes incubator org, fit that vision. I would also like us to go through and figure out loosely speaking.

C

What do we think the mapping is sink to repository for every single repository I'm happy to create the spreadsheet and iterate on this with somebody in an ad-hoc session later on, but I really feel like a few. People from this group need to go through and actually like, stop beating around the bush.

C

Let's actually look at all 78 reposts that we have and start to make this designation and through that process of discovery, figure out roughly like what the boundaries seem to be, what things fit and makes sense what things don't fit and don't make sense, and that can lead us into like whether or not it make like. If we have all these communities say gorgs. Where would things following them today? No.

E

I think the point is, the point is like I: don't want to try and retrospectively organize everything, and that's not that. That's not the purpose right like the purpose is to figure out how do we give space for people to do things as a group without giving it an official blessing and I, and that's it in my mind. That's.

D

E

Like the official picture of what fits in and what doesn't fit in, those are just positions for a different day right.

A

Well, we don't.

E

Think we should have.

A

To go through the exercise of looking at our current repose in the kubernetes org and deciding what to do with some of those. It's a matter of fairness, honestly for projects that you know in good faith, operated in the incubation process, assuming that they would hit a certain level and graduate if they just got caught on the wrong side of timing. That just doesn't seem fair. So.

F

We are going in circles again, we are just rehashing the exact same things that we discussed in the previous meeting and it is totally unproductive did. Did everybody actually read the documents it was produced? Yes,.

C

What so, what I took away from that was? Can we get consensus that saved architecture is the deciding place or what the architectural vision for kubernetes is and then, as a result, can we looked.

F

C

Like what projects fit in there, so.

F

I I could can't open your spreadsheet any okay.

D

F

Can go through and do that I also added a section to the document shortly after Phil jarred it with some examples of things that were needed or we thought were needed and the product that don't strictly fall into the core definition of kubernetes. So there are some practical considerations as well. I do think some group needs to decide. You know where the line is and I think it's personally good. It's mostly pretty obvious where the line is for most things. Some things are on the edge yeah.

J

And that's flat so for the tryina I feel like I. Don't have a rampant disagreement with the SIG's, get proposed new repos and get approval from sig architecture section like that. That is completely sensible and I feel like we could move forward on that. Maybe a little bit detail but I think at this point like we could. We could put a rubber stamp on that and have some measure of success.

J

I hope, I think we're iterating a lot just on the last two proposals here, whether it's sig bound or just one big bucket and I, don't actually see a major difference between the sync bound versus one big bucket honestly.

J

F

Practically practically speaking, one org is not going to work whether there are six specific orgs or whether they're even finer, Greenworks I don't care, but it can't just be one at work that is not manageable, like I personally am NOT gonna answer black requests to add, do random stuff to the org write like we need a whole team of people just to maintain the org. If we have one org, if you look at projects of compliment size and scope, node has a hundred repos Cloud Foundry has 300 repos right like.

F

J

That's fair I guess then: maybe we don't go and revisit all of the existing kubernetes star repos, but I do think. There's merit to what Erin was saying about actually looking at the existing incubator repos to see where they would fit and if they would align nicely with well. This thing.

F

J

Is a playground model ultimately.

F

We are gonna need every bit of code assigned to some sig right, so look every directory under package and kubernetes kubernetes every repo. Our goal is to make it clear which sig is responsible for everybody filmed in the project, so I think that's totally reasonable to go with the current repos and say well.

F

Is this a mono repo like kubernetes, kubernetes and kubernetes can trim the multiple stings are playing in or is it a sick, specific repo I think those are the two categories we have now and I do believe that other than contribute, Nettie's on and Doc's I think almost everything else, Falls and with your mistake.

A

F

I, just common.

A

Who owns those like? You know all the utilities stuff? We have to figure that we.

F

So there are a couple of unknown things like the build system and utility libraries and a couple of other things. Those are on our backlog, honestly of to assign ownership of those things.

G

So you said that we don't want a single org I'm. Sorry you for having an org for each say, so he missed that part. Well,.

F

If you look at API machinery, I think they're gonna need several orgs. One org is not even gonna be enough. We have one for clients, we're gonna need one I think for other bits of API machinery at minimum, so I'm really having trouble right. Why do you think we need so.

A

Many different orgs, what's the main driver there, the.

F

Driver a lot of it is permissions.

F

There's also some other issues with discovery, and things like that, like you, can only have six ten repos but I.

A

Mean we keep the shine, kubernetes dot io that highlights all of the different repos and what they it's not like, I'm, really struggling to figure out. Why? Because you know, there's there's companies out there that have like huge portfolios and they all bring within one organization, also I. Think.

E

That I think that the question that they could relieve it for Brian I totally get the ad-hoc problems. That could happen. I think. Maybe we just need to streamline the process more so that it's like you know, you don't get to just sort of ad hoc email people to add something to the organization or add a person, the organization you put it in a Trello and somebody runs through it once a week or whatever it happens, to be like I. Don't know that well,.

F

We also have automation and BOTS that need admin, access to orgs and things like that. That's arguing for Less orgs, no I mean unless well.

E

I think it's concerned, it's, like you, end up with the union of all BOTS that you want on the org. If you as.

C

A newcomer to the kubernetes kinney fragmenting all of this stuff across many many organizations is going to make it very difficult. I understand where the thing is that I'm looking for and where I can most effectively contribute and I really.

D

C

If we focus on fragmenting out for architectural reasons before we have in place something to solve that disability problem, we're gonna be hurting I.

E

Actually I, actually totally I would disagree with that entirely right because, like for example, I think the kubernetes client breakout has been quite successful and actually makes it dramatically easier to go and contribute to the Java client and then to then it would be if it were a minute.

J

We have data on how well attended each cig is, and what the roster list of each cig would be to know. If what we're even discussing is something that those groups could manage so.

F

Here's an example: we kubernetes used to be part of the Google cloud platform, github org, and we did not have any admins on that org. So we could not add members to the org right. We are back in that same situation like helm, adds all contributors as external collaborators, instead of trying to add them as members to the org, because it's such nes to manage members of the org. But the time were part of the sig apps sort of organization.

A

That's one thing right, so that's again, there's.

F

A

Org I mean the other option is, is you know and I signed up for this and didn't follow through? So that's on me, but we could. We could actually sort of build BOTS around managing membership lists next that through PRS, which.

F

Would be better yeah, it doesn't solve the bots permission problem, but it at least salty.

A

Pots are, we can mean, like we're, concentrating.

F

Everything I'm crouched there very few. What.

A

J

F

Staging copying all kinds of other junk yeah.

C

They're, not people actively contributing to prowl to hit this and we are at the 3-minute Park. So what I want to understand is what do we feel like we're, gonna, walk away with to push this discussion forward into actual things all right can I look to you to like put together a proposal for how people house it should try to seek architecture and what the process should look like to OHS repose into the Cure Nettie's or yes, how.

E

About we have it as a concrete since I think we all agree on it. Let's do the current incubator to like, let's, let's do the split and that's let's put in the ones we think are absolutely going to be in. Let's kick out the ones we think are absolutely gonna kick out and let's tell the ones on the line that they should need to come to city architecture. I would.

C

Rather, have a discussion over spreadsheet, we can have this in public if you like, but I, don't want to make any actionable decisions until we've had a chance to just take your rounds like.

E

Oh yeah, no, no I'm, not trying I'm, not suggesting we do this I. What I'm suggesting is that we that that's, that should be our action item. Okay and then.

D

E

We actually all effectively agree that we need to do that right.

I

Would you get like that on.

E

What does take that on mean? Take.

C

That on means, can you put together some sort of document that lists out all the things, and can you schedule a meeting to see who's interested like actually walk through that or.

D

C

Together your strong in and send out the straw man to the mailing list, so we can get some discussion on like do. We have consensus on what I.

E

Think we have to have everybody vote. I guess maybe well, okay, so here's what I would I would not not area in the community. Everyone on the steering committee to vote I would.

A

I would be game for a meeting where we go through the spreadsheet that Aaron put together look for categories patterns, clean lines what's on the bubble and try and actually derive from that. What we think some of the some of the the guardrails are. We.

F

Are undoing what I thought was progress from the previous meeting in this meeting and priority-setting? Can you help me define what.

C

Would be a better priority, Incheon Keable action Brian on.

F

Making sure that or Gantt efforts and the projects can actually move forward like moving stuff out or the kubernetes kubernetes repo moving stuff out of kubernetes contributor cat 6 needs to unblocking those things as opposed to addressing Crupp. That's already an incubator or projects that want to can.

C

F

C

E

Hey hey, can we just have a vote effectively because I think those are two totally valid paths forward right? We can either say we want to clean app incubator or we can say we want to give people path out of Cuba Native communities. Can we vote two prior, which one we prioritize, because I actually don't care I, see either way.

L

I do at the same time, no.

E

J

Viewed the action on that behind and sign up for which was this is how you go, engage with stake architecture as the thing that defines how you get out of cube, cube, I, don't know what more we would have to vote on for that right now. It seems like there's your pretty item on that and.

E

That's why I'm happy I'm happy to put that out there for people there.

A

Is the question of projects that are incubator now wondering what their future is? Why are some projects that have graduated like? Are they going to have the same opportunities that those projects have I think there's there's community equity issue? That's in front of us there yeah yeah.

J

I would like us to make forward progress on that I just wanted to I didn't think that we were going backwards. Brian I felt like there was the action. M-Net aaron said that you would write up how to engage, take architecture and I felt like that was pretty not yeah.

J

The list and getting a group of folks together to find the the word you use job it was a good were like the rails, and the patterns is, is what I think of subsets of us should go and do before the next meeting I also.

G

Want a third thing which is like to figure out what we like a vote on the proposal that Brendan had laid out earlier on that it seems, like some people were on board for in nine, six would have their own works like.

J

That gets written up a little more crisply like we're added as an option on to Phil's doc. I think.

J

G

Is and yeah I don't understand why I was vetoed because I was all for that when we talked about it, hmm can they I'm beat oh that and go.

J

That's some veto that I'm sorry I missed that in the highlighting.

C

So I will I'll take the action able to put that spreadsheet out on the incubator repost, and we can just have a discussion. All I'm looking for is to do the exercise of walking through all the repos and seeing like, what's sake, do we think, owns them and like wherever they go, I'm not trying to make any decisions based on this I just want to see what we're talking about. When we talk about these 20-plus repos, an incubator, okay,.

F

So, just quickly on the on the veto issue, the main issue was: who would get you decide what repos get created under all the SIG's? That was the main issue that there's no decision structure in place for individual SIG's.

J

Can you defer that decision structure and like we say this is the we see the process moving forward but and then.

F

Is that we defer that so anyway, go look at section.

C

Okay thanks: everybody I'll see you offline.