Kubernetes Community, 14 Jul 2016

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Kubernetes Community Meeting 20160714

Description

We have PUBLIC and RECORDED weekly video meetings every Thursday at 10am US Pacific Time.

https://docs.google.com/document/d/1VQDIAB0OqiSjIHI8AWMvSdceWhnz56jNpZrLs6o7NJY

Demo MiniKube; Demo Henge; Elders Proposal; 1.4 release features; Renaming PetSet; SIG-Testing; SIG-Scale.

A

Hi, so this is the eternities community meeting for thursday july fourteenth and we are, as always doing, a recorded community meeting. So this is public recorded all that, and this morning we have a couple of demos. We have mini cooper with aaron prindle, hang henge. Hang you sitting.

B

A

A with suraj deshmukh and then we'll go through and talk about an elders proposal which I posted this morning we have some 1.2 feature: release updates from a. We also have a conversation about the potential renaming of pet set and then the usual cig reports and notices as we have some space. So let's get started so Aaron. Are you going to pick us off with your first demo.

A

Okay, Aaron. We can't hear you yet.

A

Maybe not from Aaron now he can't seem to hear anything either easier.

C

A

Muted, all right, let's find Aaron.

A

And Dan alright.

D

A

Hey Luke: we can work on this and Suraj. How about you go ahead and do the second demo first and we will see we will see if we can get the audio set in the background, while you're doing your demo, you.

B

Don't know promo.

B

So should I start yes,.

A

B

So yeah, hello, guys there'll, be demoing, hang gay, so about me. I work at credit as a developer's engineering work on hangar and nearly killed / atomica. So what is hangar hangar is an artifact converter. It takes dr. Campos file and spits out covin. It is an open shift artifact. So you lost like why hindi.

B

So we observed that like getting started with cubanita an open ships is hard because, like you have to know all the I mean at least some basic concepts to get started with and write the artifacts and get your app into working, but then for an app app developer. They cannot learn, learn the concepts at the same time and also get everything right and deploy application. So so, if you look at so why we chose dr. Campos because it should look at it. It's it's. You I, its user interfaces like easy to get start for somebody.

B

The curse reps is here: yeah I mean come on. It is a solving the American engineering problems, but I, don't containers but but learning constructs is, is not easy job either, so so yeah. So installing hangar, it's it's very easy, go get and certain Ingo usage, some of these commands I, will show you in the I'll, show you in the upcoming demo, so yeah alright.

B

So we have a docker compose file which has like two services running here: n we're on 16.

B

B

B

So basically, it converts everything reading the doctor compass file, whatever is required for running that application.

B

So apart from this, it also has the interactive mode where, if you don't provide a poll vote for particular pompous office, then if you like ask you and then you can, you can provide the port number and then create service, for that also, here, like you, can see so also also, it works with the open shift and similarly, okay.

E

B

It and see so so so are we. Our way forward is like so we started out with open shift, only so often shift cam with the feature of importing taco compost file, which which took the napkin postcode and I used it in open shift internal. So we took it. We took it into hangar and created some common line FS around it recently added community support, but then recently like lost last week, we saw that like compose booting into into compost. To recently we had talked talked with them and we were viewed.

B

We are willing to like collaborate with them. We would like to have something that is a canonical converter, which makes and merging the efforts and everyone works on a single thing, also yeah. So so main intention of this demo today was to show that we are interested into this space and and yeah.

B

We would like to work with the community to get this thing easier for everyone, so the main thing is to bring bring more people to Copernicus and and and in effect, tiny as a side effect will end up, adding more things to communication, open ship and with skybox composed too. We have already started adding creating issues and collaborating with them so yeah, that's it from myself and wine for the links you can. The links at the end of the slide I will share this slide with server so that she can put it in the doc.

A

Thanks Suraj: does anybody have any questions for him.

A

Not this time we can reach, they can reach out to you. There will be notes in the dock, with your slides. All of that. Okay, let's see if Aaron and Dan are set up, they were pretty close. We heard Aaron for a moment there, sir I'm, going to mute you and let's see if we can get the these guys going.

A

A

F

Right, sir, oh yeah.

G

That's an echo yes.

A

We can hear you is.

F

It yeah dear me: yes.

A

We are so great.

F

Okay, awesome, alright sure yeah, so, okay, all right, so uh alright, so on your infernal I'm with Dan right now and a part of the cloud developer, velocity team at Google and today we're gonna be showing a mini cube. So what is make you mini cube is a local development environment for Cooper Nettie's, and uh essentially it has everything you need to get started with Cooper Nettie's and rather than just yeah, can we talk about it? I'm gonna show you them home.

F

So right now we have our mini cute binary, which I got here from our releases page everyone's just installed a mini cube and Iran mini coupe start. It took about a minute 30 seconds to get started. I've configured um soccer, the two pretty much use in our committee cube and let's get into it. So first thing: I'm going to run we're gonna set up a proxy to visualizer using cute control.

F

uh So now what I've done is pretty much I've set up like a visualizer, so you can kind of see what's going on and now I'm going to run a deployment using cube control a little so many cubes running the background. Its provision to vm, that is everything- that's needed, uppercu benetti's.

F

So now I'm going to supplement and there's a page and you'll see here at essentially, I started a pod in communities, and this is what we're seeing right here through mini cube, and what this looks like architectural e is that so we have the mini cube binary, uh it went and it spawned a vm using a lip machine, and so this DM I started running a local cube, which is a Oh binary that runs everything. That's necessary, Cooper Nettie's. So in having a mini cube, configure your tube control to talk with local cube.

F

We can now use cute control, just like you would in your cluster ordinarily. So we can do something like Yale up our pods and we'll see the output in the visualizer. So this will get scaled up the four pods and we can even do so. If you just kind of iterative development, we can edit this we can edit our pods.

F

And I would see here that disappointment, add replication controller in it that will go in and change our Nautilus pods over to kitten pods. So and now what you through mini cube, we have full access to the ladies API, um just like you want, and so this is kind of like a like a basic demo with this visualizer, but you can also now you can kind of think of now that we're running locally.

F

You can actually be more previous than previously, because you don't actually even need to upload images like our docker images to repositories, we can just run local images all right now. I'll use a mini, Cuba feature this dashboard, which pretty much just accesses a dashboard so that you can see um pods as they get provision and then so I mentioned its prior, but so I'm just going to show you I've pre-built, these two hello world images and I'm. Just going to show you hear that these are all the images accessible on the vm right now.

F

So we have our hello world image and what this is is the note app that's going to say hello world, so I'm just going to uh this is our deployment for this app. So you can imagine so one who's trying to global development. They made a disappointment and these images so now let's go and create this oops.

F

Okay, so it'll actually show up in the dashboard.

F

Okay, so we see here that we provision our this, uh like 30 replicas of our note app and whether we can access this is now that we are using a mini cube. We can use a service, so we use mini, keep service front end.

F

Oh sorry, I forgot to know we need to expose our the deployment we just made so I'm going to make a service called front end that will expose it and then we using the cube service in the name of the service, and this will actually open up through the vm using a proxy like we can we'll open it through that note or service that we made and we can actually access it. So now you might imagine someone to try a new some kind of local development. Okay, instead of this hello world app.

F

Maybe they want to change the text or something. So we have an updated version here. I've already built this image, as you saw on docker images, um so we'll go and we will edit our deployment okay and this actually. uh So this, like is going to take like brother like roughly 30 seconds, but so essentially you can see here that, through mini cube, we've pretty much. We started having to dr.

F

build soccer, push to GC our soccer, pull that down again, um so we kind of like made that wonderful passer also if people aren't comfortable trying the images that they don't know where they want to take your images.

F

People can now, instead of having to figure out how to use a private repository with your benetti's or update all the api's, they can now I'm kind of just do local development, which we're seeing from users is it's like a boon for a lot of people, so just going to make sure okay and we have a version two of our app okay. um So so this is just kind of way that we're kind of hoping that needy cube just trying to allow local development for cuva.

F

Ladies users, to be kind of quicker and easier to get started. So just from the main page of eq we have like. So the main purpose, as mentioned before mini cube, is to make it as simple as possible to run communities from your local machine, and we can see here some of the goals that we have listed so right now mini cube, has a couple dependencies.

F

For example, you saw that provision VirtualBox machine in my case I, currently sports VirtualBox, vmware fusion, k, vm x, I've on Mac, um so we're trying to make it so people can simply download mini cube, and that's the only thing that any of you started with Cooper Nettie's we're trying to get more feature, support on the future. Things like a post directory mounting and all the functionality communities like, for example, ingress proxies like these, like that are features that we're working on.

F

um We hope to get a rocket container support in the future, and just a little bit about the community has been a phenomenal and very supportive, both communities you can see here. This is our star graph that we have, but it's really the community. This seems to be a real pain for people.

F

Is a lot of people really want to use and develop for communities, and they they don't exactly know how to get get started or the best way to do development on communities how they can rack with the newest features, and it seemed to be a painful infrequent that the community is kind of showing that, and we appreciate all of you guys to go and check out and you can become part of our community, because we think that it's always like a real problem. A lot of users experiencing.

F

Thank you guys so much.

A

Thanks Aaron, it's great to see the engagement with it in the use cases in its case and people are seeing finding it on github and playing with it, and appreciating we have any questions from anybody. I just.

H

Second seras comment: it's awesome to see it's been a long-standing pain point and it's awesome to see.

A

E

Used it myself so.

H

It's it's cool in that sense. Thank you. Yes,.

C

It's pretty awesome, I beez jellison any.

B

Thoughts to support hyper-v as well.

F

You can hear me yes, oh yeah, oh yeah, yeah, absolutely so that is a kind of a future I think it's listed um in the future. We do hope to support pretty much like removing the the independency entirely and supporting like a hyper-v like all the built in a virtualization hardware for like Windows and Mac, but as of right now it's like more of an extended goal. Right now, that's like probably more like q4.

G

Yeah, you know, there's nothing really blocking us from the supporting hyper-v today. We just really have anybody with Windows extra fuse around you have to start working on it. Yeah is.

A

That an opportunity for the community to help you I.

H

Was can see what it make sense to target libvirt because that will enable all of I don't know if you guys are already using the vert, but if we target the vert and I think we can basically whatever it is Asian ecology. We.

C

Probably can help with hyper-v from the trend. Oh.

A

So you've got some volunteers for help and work and ideas. So how is the best way for people to reach out to you through the repo we're gonna these needy cube? Oh.

G

Yeah, that's the best way to find us right now we're targeting docker lid Michigan, actually, which supports most of these virtualization providers and one.

A

Of the drivers is.

G

Live for it, so we can kind of get the whole bunch replaced already just a matter of building and testing, for he goes.

H

You don't happy to take a stab at building and testing school now that I have a windows. Let go.

A

So much fun, alright, uh any other questions for the mini coop people.

A

Thank you, Eric man. That was awesome. It really does a space that we need count the new fruits. Thank you alright. So next up on the ladies, a gender list is an elders proposal, so I've mentioned here before that we have we, as the community we've grown very broadly and very rapidly, because we chose to have a benevolent dictator for life sort of model.

A

We've ended up with the painting and jewelry's of one of the, not so a clear and transparent any port hierarchy, but that also means that we have had the benefits of offensive of meritocracy. As there are many bits written about valve, one has talked about many a great length about how an inclusive power structure forms if there is no expert power structure.

A

So in order to make this a little bit were cleared, watchman's parent of how to grow from the communities community and how to get disputes around technical things or social things or person, personality and personnel things resolved. I have put forward a proposal in the github community issues, space that we have called the elders proposal. Basically, we make a triumvirate which will grow to be a five-person body through engagement with the elders and that these people would basically be escalation points for potential technical disputes.

A

They could act as a Supreme Court, hear arguments on both sides and then come to some sort of decision. They could act as a a feature in or out resolution point with some sort of with a similar sort of model. They can help handle any personality disputes. They can help us grow, be the contributor progression model figure out how we're going to grow the community but basic, be the highest level of technical and project vision. Leadership, which includes both technical and person, person and open source community management leadership as well.

A

So I would love to have comments on that proposal over the next week or so the big point behind it is this. First time I would like to put forward a slate of the first three and then have them.

A

Those three choose the next two to fill out this late for the next year and that will give us a a nice first year model for having others in place and people that we can escalate to, while still, whilst figuring out, how to make this a more democratic process going forward and how we want to maintain this escalation point in our community. Does anybody have questions? I know I, waffle down a bit.

I

Let me uh let me jump in by saying. First of all, thank you very much. I think this is a really huge, important thing to the health of the project. Overall, oh thanks for taking the first crack at it, and I hope everyone gives us there a full support thanks.

A

E

A quick question on the motivation behind the two-year terms, specifically so.

A

The the reason I put a two-year term is for stability, so I want I, want to change roughly half the group or or leave open to change. Roughly half the group each year, as opposed to having of the potential for a wholesale change.

E

Ok, so the group itself will have some turnover on a yearly basis, but there won't be everybody gets two years. Okay, that makes sense. Thank you. The.

A

First year was the one that's odd, because you have to set that up so I believe that we can come up with a three-person slate that is relatively uncontroversial. You know the people who are acting already as leaders we put that forward as a slave to the community. They two more but those two because they're brought in through relationships in the existing experience, as opposed to broadly from the community I, wanted to limit that since we needed to make it a rotating term anyway.

A

So the first three that will go put in as a slate with a vote from the community will be a two-year term. They will choose two people to augment that which would have a one-year term until next august and then from there every year some portion of the elders could rotate or they could make. You know if they're doing exactly the job that everyone wants. Then they just get voted back in.

A

Thank you Eric. Anyone else have questions about this or you can reach out to me. This is this is just me trying to put a straw man out so that we can come up with a nice governance mechanism for how we want to run the project as an escalation point.

I

Would you consider this one dot for roadmap item sure, just by way of targeted timeframe for having something in place, it seems like kind of on the release. Boundaries is a would be a good way to think about it. We.

A

Could do that I was hoping we could vote for august one once I'm going to put forward a slate or I would like to put forward a slate next week and I'm open to suggestions and people offering me names. I think I given that work, starting with people who have you know, been long-standing large contributors in the community. I don't think that it's going to be very controversial, but if people have specific opinions, I'd love for them to send me this way.

A

I was going to put us forward islate next week and then spend the next. That would be we'd have another just about a week for the beginning of August. In order to vote, I was trying to move it more quickly than the the end of one point for having it in place, but we could wait that long.

A

If there's, if there is further commentary on getting this started, I think defining things like who who is able to vote in next year's election and how we do a nomination process for the 2017 election would be things that could be on the 1.4 timeline.

E

Yeah I agree: that's fair I like the idea of having this council in place prior to the code slush, so that they could be available to veto or Meiko novogo decisions, for that is that sort of a right is that an intention of the use of the elders council, just so I sort of understand what we're talking about certain.

A

Certainly so, um yes- and in this case, putting forward putting forward a slate as quickly as possible, is to get us to something that is working, but also I also left in the document that they could add an additional two members or would be expected added additional two members over the next year. So.

D

A

Limit what they can do now, but it means that they would need to act and make those choices as soon as it's reasonable.

E

This sounds great thanks.

A

The voting next year would be much more formal this year, I'm hoping to be, you know, generalized consensus, I, don't I, don't expect that this will be contentious. As I said, we've got, we've got many fantastic contributors who have been long-standing with the project and are clear leaders as it is. So I think the vote this year is a little less like of a true democratic vote and a little more like yeah. That sounds cool as opposed, and you know, and if people have real concerns then then we have a longer conversation, but I don't.

A

I don't expect this, this initial slates to be contentious in any way.

A

Good enough, oh, if yes, please, if everybody's happy about this, then send me suggestions if you have particular names or obvious people that you think should be participating this or might not have been considered. I have a long list of people and have been turning to figure out the best starting dynamics for this. So if everybody is all good with this, I will bring forward a slate next week and we can then through the dev mailing list, and this meeting basically ratify the slate and then from there.

A

We will move forward and build out during the 1.4 timeline what the voting criteria are for next year's, more democratic election and then also a nomination process for next year.

A

All right sweet, please reach out to me. If you have opinions on this because I love opinions Igor, are you ready to talk about 1.4, yeah.

J

All of us ready always.

A

Ready I love that yeah.

J

uh So unfortunately, I have no ability to share my screen because I'm using my android tablet for this girl but yeah postolder links all the necessary links to web mitcham midi notes. So, let's start from from the first item, determine it I'd like to propose so I'd like to propose prayed there.

J

Wandered for milestone within the cabinet is features reform, because the current moment I see that we have a lot of features that that we may include into the following within one word for at the same time, we don't have milestone wanted for yet so I'd like to I'd like to receive some situations, saves on SN a concern, sir such like that, if no so let's go ahead and add milestone one at four and Mark all their existing feature. That going to be added to wondered for Lisa smile, milestone, wonderful.

A

Fantastic anything else we can do specifically to help in short term.

J

Yes, so I see did that sum up. Words are now it's shared forget in debt. So another item that I'd like to raise up.

J

So possibly we have to populate with some extra labels our fishes Ripper again, so we have to populate with some extra labels, as we do have that in our main rebel man can be necessary for because we expecting to receive much from of features in the nearest time while wondered for release a scope is going on, so we we have to clarify and define the existing features and swords I'm in some way, and so on so be grateful populate with some extra label, so I can have and main cabin extra cool.

D

Can ask a question about features, get sure at some point. Does it does anyone else think it'd be useful to just define what level of maturity or consensus is required before we put something in the features repo I know it's a long ways from it, but I don't want to get to 20,000 issues. You know next year because we have things in there we're not working on anyone else concerned, or is that just me.

J

Yeah I also searched about did, but they haven't thought about the way how to solve it. That's right now we have everything fine in our future through but yeah my grow is use it one day we may receive 1,000 new features from any any people that it won't work on them, and it's simply like, like some dust in our or a vicious rumor like.

I

My two cents is a much closer I think where Eric is going with this I'd like things that move into the future repo to be at least in a fairly mature proposal form and in order to get to a fairly mature proposal form theoretically with the sig or some equivalent. That's just my two cents I, you know I, don't want to slow down features and or anything like that, but it really should be four things that you know. Theoretically, will land or portions of them will land in the neck?

I

The next milestone of the master fallen.

J

Snow like oh, the features has to be approved and agreed with seeks before they little bit yeah and it's it's like the part of our float that we have already discussed equation proposal is Mike's afar off here, so yeah. That sounds good and that's why I'm also proposing to add some labels. 28 labels like six way, is marking this six groups or anything else how to define what what specific sake is going to work and what specific c-can we interact. Waste forget in this video yeah and it should be a good filter for that.

J

So, like a single, a twosome responsible people from the six-wheeled bear like responsible for this feature to video right and.

K

I think part of the proposal original proposal is features or also that I cigs will actively triage the proposal. Station touch features that attached to this particular cig right and sometimes for teachers may be merged to remove kind of closed altogether.

A

Your kit, with or and for your commentary, David.

I

Yeah, that's fine, I agree.

J

Yeah so another item that would like to the day I have already done in my personal for cough features. Repo I have moved the wiki page list. It describes our wonderful mouse on a road map from community report from community wiki to feature cerebral as like classic readme file like.

E

J

File so I post a link to it in in a meeting notes again and possibly it should be a better way to well like to to track and to find some information or getting our existing religious, so well update was there from Lane yeah, so I supposed it should be better to to have these items in our future strip of that community week. Cuz community wiki is a bit front place for that. In my opinion, you.

A

Are dims asked about the 1.4 release schedule where that is yes,.

J

I know I noticed it and deems you might find their release kiddo in a in a major notes. I have already posted a link to due to a file that describes dit Hank experience, I'm speaking about it, that mugged on file that I moved from wikispace from commentary, but honestly.

A

That explains why you couldn't find it. Thank you. Yeah.

J

So sorry simply approve my suggestion in intimate inner. Imagine else because I I have no rights.

L

To edit page, he or what I was implying was that we should publish, publish it somewhere where people can find it.

J

What do you mean? What would what place you may propose? Is not a google doc I.

A

Think that I think that there is a published, 1.4 and 1.5 timeline that TJ wrote up that I need to go, find so dimon's. I will try and find that, for you did.

J

Did you send melon in a melon lesen cabinet as deaf and scare? Oh, and this information is also included into community we get and to markdown file into a future circle? So thank I supposed it's enough.

A

Awesome all right anything more on 1.4 features.

J

Yes, so another another question and as a two main question, so that I have so, are we going to move our already proposed features to uh to Cuba necessary po2 to the new feature strip? So I suppose, if we are going to include some of your features to wander for relay Center to the fuser federal releases, I would have to move some some issues and sir pr's from cabinet history, but main concern that I have that's how to move discussions. So some issues and Paris April a huge discussions and Errol important for I.

D

J

D

Some thoughts on that yeah they should not be moved. The features repo should try to be a summary and discussions that are legacy in the coup. Benetti's repo can stay there. They can get turned into proposals or design docs or whatever. That would be great. That's part of features we go enforces the discussions get summarized in the future.

D

Going forward of people want to use Google Docs and that their sig that's what their sig wants to use for discussions or email or paper airplanes, with notes on them like I, don't care as long as they're, not in the features, repo, every single +1 and what? If we did it this way, because I want features repo to be a place where people that have to have a broad view can go and get a summary.

J

Okay, so we may simply not move but duplicate. The proposed features for the new laser say in the future three pi and simply condom se as a peer features using the issue template it did you proposed.

D

Totally expected that you have a features, repo issue which summarizes like the lifetime and all the lifetime evolution issues of it, and then it links back to here's our initial discussion and that's a coronary po item, no editing needed for either you just have to parallel ones and they have different audiences. Okay,.

J

Okay, I get it yeah and another question. So what's what we're going to do with some non-core cabinet is features, some not features but items yet like see. It worries some extra installs that have their own repose in communities organization, but they are like early important as well for for the wall. Cabinet is as a product, so I suppose we have to.

J

We have to include some extra features for for these components, to the features reply as well, so like track them in a in the in the same percent in in the same some vegan, yeah, yeah and yeah. The final item from me so I am constantly populating the spreadsheet it that had been suppose really useful for a foreign in cabinet is one two three a few weeks ago, so I'm going to so I'm populating it right now.

J

We swap it for peaches so and also I am expecting to duplicate it as a knockdown file in which is rep as man. The link is also based in to him to our head, announce.

A

Your awesome yourself documenting your meeting. That's great! Thank you.

J

Yeah, that's all from the impression awesome.

A

Thank You Igor and thank you for all of this work against 1.4 milestones in the future. Repo and digging in on this I know you and Mike stepped up at the end of 1.3 to sort of figure out where we were with all the features and then give us a better summer. Even we had had three weeks prior to that, and now getting us on a process for 1.4 is awesome. Thank you and Eric. Your features, repo is going to prove to be invaluable, I.

A

Think, okay, next up on our agenda, unless anyone has questions for a no more questions. Okay, next up on our agenda is a pet set David. You said you had an item about discussing pet set and changing the name for it. So.

I

As per the standard for computing, there are only four hard problems in computing: I ruined the joke, three hard problems in computing, naming cache invalidation and off by one errors. Of course, yeah. The naming is one of them, and in this case in particular, we don't want to make this. A big thing we cut set was always intended to be a code name. It's not a thanks, claim. I know I totally ruined it. Sorry, moving on and Justin. Sorry, I will stop looking at chat. Uber Nettie's is not a name just to be clear everyone.

I

It is a code name by itself. The and the answer is its cross cluster Federation. That's it okay, so Luba Nettie's is a codename pen set was also intended to be a code name. It is not follow our descriptive model of how we like to name things in increments, it's a little bit too cute and so we'd really like to move to a more descriptive name. We do not want this to go on for some insane amount of time. You'd like to close this very quickly. We may end up and landing on pet set.

I

Simply because people know it, but that said, there's a give hub issue open. We would really like your opinions way in. We have some good candidates right now and we'd like to nail this in the next work you, but that's basically it what.

A

Are the what are the good candidate names, heinous.

D

There's like way more candidates and our people in the discussion.

A

D

Like each person adopted in most one that they wanted to yeah speak for that would be a nice way to narrow it down. Yeah and.

I

I mean I think way, I won't. I won't give my bias right now, but I so I won't name any in particular, I think we have some very clean ones that people are already starting to coalesce around, that are nice. They are descriptive and- and we could settle on this quickly, my goal would be like as soon as the end of next week, we've all coalesced around a name, because there will be obviously some changes in code and and grunt worker involved here.

I

But today's I say we would like you know we're opening the door for a very short amount of time to rename it and this door will close soon. Yes,.

A

Putting your votes, okay, so everything is going into issue. 27, 430,.

I

Yeah shoot I meant to have that up. Thank you can even link that. Oh there is.

A

Good, yes, in fact, it wasn't me who did that it was a community member who I already said.

C

Yeah siraj. Thank you. Thank.

A

You, sir us, yes, fantastic! Okay, uh then we are on to the sig reports section. Unless anybody has questions.

A

Okay, where we've digressed into chat I'll, stop reading this? Okay, um any other. So.

E

I did just have a quick question. I guess it seems like we encountered this once before, with replication controller and replica sets, although replica sets include a slightly different concept and replication controller. In that case, we didn't rename the existing thing. We made a new thing. Are we going to do the same thing here with pet set, or are we going to try and renamed the existing thing? I think.

C

Walt will start remains to be determined, but the where we did with replica set um lives done because of a difference in semantics, but it actually is causing some pain. So we may end up trying to make it compatible with replication controller.

C

hmm Yes, is the existing object beta, which existing office the.

D

Existing pets at all.

I

Well, that is all absolutely no commitments to anything. Yeah.

H

In that case, I I would vote for just renaming the existing one. I think that it, because when I can change the object itself, there's not a strong reason not to just alias the resource pads for.

C

Some amount of time for.

H

Some amount of time yeah at least the duration of the beta, or something like that. So.

C

Anyway, we can discuss that on the issue if we decide to rename it yes.

A

Okay, awesome any other questions.

A

Okay, any cigs have things to report out, I'm going to start reaching out to cigs one by one and saying it's your turn to report. If we don't have more jumping up I.

E

Can call out what we discussed during cig testing most recently, and that's just that employments testing. Is this the kind of a thing that doesn't work terribly well, if you're not using the existing cluster, stand up code. We're encountering this specifically in the context of their mettle testing.

E

There are a lot of things that assume SSH access there, a lot of in dentists that assume SSH access to either the master node, and then it assumes the master node has a certain set of components living there or SSH access to the other nodes. This is currently tied up with the concept of a cloud provider. So, if you use AWS, your key needs to be where a wsss underscore SSH underscore key tells it to be, if you're using google cloud and something different, we're thinking. This is a little weird.

E

It probably makes sense to either create the concept of a bare-metal provider where SSH keys can be put in known locations, it's sort of like a deaf or starting more like a nail or null provider, or maybe we just untie the concept of ssh access from providers entirely and pass that explicitly to the e2e best binary. But at the moment it's really tough for somebody who's standing up a bare metal cluster to actually verify that their thing is conformant.

E

They have to like tell the tests that they're using some cloud providers so generally folks have chosen to use AWS or something that doesn't have too many constraints like. If you don't set a cloud provider, it assumes google cloud by default and then it seems you have g-cloud installed anyway. These are the sorts of things where you've been discussing on a week by week basis at sig testing.

E

We've also been looking at trying to get the testing for the entire project off of Jenkins and onto a Cuban Eddie's cluster I think there's been a lot of fantastic blur work on the test grid. It continues to improve on a day-by-day basis, and human teddy nadir is now at the point where, if you have a failure on a PR, you can click through and it shows you the exact indication you can use to rerun the test.

E

The exact failure that happened in the code that you can go, look at to see what, when we're home.

A

That's huge step so.

E

Anyway, lots of fantastic work going on cig testing, weekly Tuesdays, 930 a.m. pacific time.

A

Excellent, thank you so much hard. Alright, anyone else want to do a sick report. Oh I'll.

I

Be happy to say a couple of words: okay about six scale, a quick admin note we've started recording as a last week we started recording the sessions and we haven't actually post them to a YouTube channel Sarah. You and I've been working out how to do that. So I think we'll, hopefully get those content where you can see it. We have been continuing to try to take pretty good, didn't if you're interested.

I

Please go over and take the notes, but from recently I think the topics have been the group's actually been a little bit slow across the 13 release. Time frame, but today we're kind of back to our usual level of interaction.

I

One thing that's been going on across a few sessions is we're continuing to work on how to do performance performance run, exchanged performance data exchange in a reasonable way. It's really hard to exchange data and a standard way to let people compare performance results. So that's been kind of long-running thread that we're at we need to work on.

I

There was a little fair bit of discussion today, for example about 1 dot, 4 and what's going on and wound up for, I think it aside from some scalability some throughput improvements that Google is working on wound up for is lining up to be mostly sort of stability, ketchup release as opposed to expecting any big, any big things in it.

I

So I guess the last thing there I mentioned is there what the nature of the sixth scale group is is such that it tends to be a little bit kind of broadly sweeping across the whole project, because you're never really sure what part of the system is going to be the most influential in terms of total system performance. So we frequently end up talking about various topics that we wonder what the impact will be.

I

I think the one that was a biggest note from this morning was the the new big there's, a pretty sizable trunki Intel proposal, which I think after discussion we concluded was probably not that impactful to overall system scalability, but it's large and interesting enough. That I think the request would be. Can we get in till to give us a short community proposal, presentation on on what they're thinking about line so that'll do for today, I think fantastic.

A

Thanks Bob, Brian and I are also meeting with intel on friday to find out where we're in ways that they want to help us so that we will tie that back all right. Any other cigs don't want to step up.

A

Okay, I will tapping shoulders for next week. A last thing up, then, on my agenda or notices. The second half of the 1.3 community retrospective is tomorrow, 10 a.m.

A

if you're interested in it Jason singer de Mars from Deus is going to be leading it again and if you're interested in participating in the online karunanidhi's meet up and the organization of that, then a and Julian and Sebastian are working on that the first one should be the week the last week of July and the idea with that one line made up is to be focused more on on how users use and learn Cooper Nettie's it'll be a lot like and real live meet up, and here are cool new things only online, as you might imagine, okay, anyone else have any other notices all right unless you were on mute, we're going to end up ten minutes early.

A

So, thank you all for joining, and this was actually incredibly productive, because I didn't think we would get through all the agenda items. So thank you. Alright, see you guys next week. Next.

E

Week, thanks for ending early Yeah right ten minutes anytime.

L

A

The gift of time.

A