Kubernetes SIG Contributor Experience, 15 Feb 2023

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: [SIG ContribEx] Weekly Meeting for 20230215

Description

[SIG ContribEx] Weekly Meeting for 20230215

A

Hello, everyone and welcome to the second February meeting for sick contributor experience. um This is a general reminder. These calls will be recorded and posted sometime on YouTube later, as well as we abide by the cncf code of conduct, which essentially boils down to please be excellent to each other.

A

um I have oh, actually, I did not drop a link to the notes in chat.

A

If you could, please add your uh named list of attendees there um and with that we can get right into our regular, scheduled business. uh No I do want to give a quick update on the summit.

B

uh Sure someone is going along smoothly. uh Registration has launched officially Chris, so kindly sent out an email this morning. uh Everyone should have the links that they need. uh Cfp will be coming soon. Hopefully, within the next week, oh cool.

C

uh uh I will be on vacation, but I will not turn off the the kubernetes slack. How about that? If you need me next week.

A

If you do want to turn off career slack like I, think I have your phone well yeah I have your phone number and signal, and things like that, so.

C

That's true, yeah I mean anybody could call me yeah, you're, right and I. Think I even put my phone number in my profile just in case yes, I do have it. Okay, so yeah I.

A

B

C

B

I parsed that, as you weren't, going to turn off the entire kubernetes slack before you leave for vacation, which.

C

uh No, what I normally end up doing is deleting the whole app I literally delete it, and then I set it back up when I get back smart yeah.

A

I do I do have one recommendation is change your name to explicitly state that you're out of office, because no one pays attention to your status. No one God, it's like what like, whenever I am out, it's like I, missed, probably tables Dash, no of it do not ping out of office until you know so and so date that doesn't stop people from pinging me, um but at least reduces it from like. You know, coming back and having like 99 little red dots.

C

A

Know less than 10. good.

C

Point I will update that um I have something else for this meeting, but completely unrelated to what we're talking about right now. So I'll save it.

A

Okay, well, honestly, I think this will be a pretty late meeting to be with. um Is there any other Summit related things.

C

uh Oh, a tweet I did y'all look at that sorry, I put a tweet in slack and I. Don't know.

B

And- and speaking of that, did you toot at some point.

C

I will toot that's easy, but I wanted to do the same thing. Does that make sense? Yes, okay, so let me actually.

B

Let us know so: we can boost your toots. Yes,.

C

Which is just fun to say: Trudy McGee, son of staff.

D

Does that mean that we have a mastodon now, no.

C

We'll just do it through our own kind.

D

Of deal oh personal, okay, I was like did I miss something very large, no.

C

Not yet do do we have any updates on that Bob. Just out of curiosity, no.

A

um And I would still probably hold off on doing anything until there is an official thing.

C

A

No, no! No! That's.

D

What I was waiting.

C

For was like word on the official thing.

C

So no news is good news. I guess at this point.

B

Okay, any questions.

A

D

I actually do have a question if I could ask since I have the opportunity- and it seems like it'll- be a light meeting. um I do plan to be there on site and I want to volunteer at the summit. um How should I, let folks know about that.

D

C

Just let him know you want to be a shadow or.

D

Helper, whatever, whatever you want to call it, your.

A

Gopher a golfer like go for it.

D

C

Really got Hands-On on the community was I. Think the the Seattle Summit is where I volunteered and like I was.

D

C

And yeah off we often here we are yeah.

D

Always help happy happy to do that and.

A

And gopher you know works for kubernetes because it's a go thing and we have you.

D

A

It's all about being on brand.

D

Cool that was my question.

B

Also gopher was on The Love, Boat and kubernetes nautical themed, and so we're coming back around again with the secondary. No nobody coming with me. Okay,.

C

I might be a little young for that uh I mean I, know of The, Love Boat I, don't recall ever watching.

A

Yeah I might have caught like a rerun here or there when I was like sick and home.

A

Foreign, that's probably it.

A

Okay, uh well, we will move over to mentoring. Is there any updates on um actually I do have a well no, never mind. I do not have an update on on mentoring.

A

That's for when I can turn off the camera.

D

A

um So I guess that takes us over to comms.

C

D

We Exist we're.

C

Still having meetings We Exist, we still have meetings we're not having a meeting this week, because both Chasm and I are out um we're.

D

Doing a lot of a few new folks which is exciting. Yes,.

C

We're doing a lot of troubleshooting around Twitter Twitter in general and true tweet together, uh the zapier account is probably going to be. The solution we need is what is looking more and more likely so for for what.

A

It's worth like that's on Jeff's task list when he gets back from vacation fair enough next.

C

Week, okay, you put that on there cool, so I'll I'll leave the ticket alone for now and just be like responsive, um easy enough. The.

C

What was what was the other thing? Oh we're. We've got we're gonna re- Embrace project boards, so that we can better track stuff in flight and people can pick things up easier kind of deal because um our.

D

C

Method is just a huge table in our notes file and that's just not sustainable.

D

It was lovely for a smaller group where we had items that we just needed to track, but now we have enough people and enough items that we actually need to prioritize. So we need something that can help us do that. So moving back to project boards.

A

Something else that might help is at this point: we do have some automation of labels, so things are labeled with contributor columns. It could potentially automatically be added to the board that.

C

Would be dope, where does that get set up at labels.

A

No, no! It's it's a proud job! Oh okay! So essentially it's just like sitting there like when a new issue or PR comes in and it has a label. It knows to you like basically call the project boards API and edit on there nice we're doing it for like enhancements and a few other things right now.

A

Every Cliff that is something you're potentially interested in. That is an option that could be explored.

C

Yeah, that would be great it'd, be great to have everything just show up in a backlog until we triage.

D

um Agreed that also solves one of the other problems that was on our chess board, which was making sure that we know about things that are happening so potentially yeah.

A

As long as it's properly labeled.

C

Right long as it's properly labeled, all right, I just sent the Tweet out for Kate's contributors about the registration being open.

D

C

Getting a linky I'll be linked. Tweet um drop that in Summit staff.

C

Don't let me copy pasta it over to 2D place.

C

um I have a question and it I might be completely derailing us. The there's. A new cluster API provider for cloud stack folks inside are like Hey. How do we set up a meeting? How do we get owners? And this new thing we just donated that you know so we can actually work on things and it's like a brand new thing. So it's not like people have like stacks of experience already in the kubernetes project.

C

Dems is currently an owner of it right now, but that's not sustainable going forward and you know getting a meeting on the books was something that someone asked me for help with and I was like. I, don't even know how to create a meeting other than ask Bob right. It's.

A

um The Sig leads create meetings for all their sub projects, Okay.

C

Cool, so I was right to bug the life cycle. Channel.

A

Yep, the the one problem with sick cluster life cycle is that their mailing list is so large right. That uh counter invites don't actually go out to it anymore.

C

hmm Yeah I mean this is one of those things where it's like they went to. The contributor's site saw the calendar and, like oh, hey, can I get a meeting on here, because I saw another provider with a meeting, um so yeah, okay, I'll I'll, poke sick legs. Sig leads as far as the ownership stuff goes. I, don't know how that's gonna work out.

A

um And that is a kubernetes six project right, yeah.

C

Yeah yeah, it's already been migrated over and everything. So that's why they're asking right like they're asking questions like hey? We want branches and stuff and I'm like oh I. Don't know how to do that. Yeah, like I and I, know I have admin privileges on some repo I could not find it to save my life.

A

um Remember all admin permissions and things like that are tracked in case.org. That's.

C

A

Okay- um and we have the two groups for every repo, admins and maintainers maintainers is Right access, admins has admin access, um but if they are messing with Branch protection and things like that, uh they should look at setting that up in prow.

C

Yeah, that's what I told them I'm like there's a proud job for this here. The link.

D

um But it wasn't Branch.

C

Protection, it was like release versioning stuff that they were thinking. I, don't know it might just be. They need to adapt their practices to the way we do things too. I didn't roll that out.

C

Cool. Thank you. I appreciate that.

A

um Prepared for all sub projects, that's pretty much like. If you have a question, it's you ask your Sig lead.

C

There you go cool easy enough.

D

C

D

The notes I'm gonna record that, as like calendar stuff talk to Sig, leads yeah.

C

Exactly sorry, um I was just looking at contributor docs and the contributor guide. Since it's the next thing, I did notice. There's content on the like contributor site is that old or new. The.

A

So the the the the um course is new.

D

A

The guide and everything else um is ingested from the k, Community repo. Every time the site builds.

C

Got it? Okay, that's right! I! Remember that now see about to go on vacation. My brain's already uh got it cool. Sorry about that.

D

And as far as calms go, I think we covered everything important.

A

Does anyone else have any questions comments uh for comms.

D

uh We there was one other thing: I was going to say, which was um we were saying that I should probably reach out to the cncf help desk to just like have another Point watching the Mastodon thing. Instead of just relying on you, Bob.

C

I feel like that's, going.

A

To be we'll we'll talk about that later, yeah.

C

A

C

There's probably going to be some wide guidance issued once they do make a call yeah.

A

um Cool okay, um that's.

D

Another that's.

A

Another uh not recording thing.

D

A

um Does anyone have any updates regarding contributor, Docs.

A

Okay, the probably the one piece news, I, don't know if it actually got covered in the last meeting- was the fact that we launched the course so.

D

C

So, oh really I think I missed that memo. Yeah.

A

Same so, if you've got a kits at Dev, slash course, uh it is the new contributor Workshop, essentially in self-paced form. We.

C

Need there are some.

A

Things there are some things that probably do need to be updated, but.

C

Oh yeah, yeah yeah.

A

um And then the the other thing I would actually probably hold off on tweeting, um because Craig is going to um get all the stuff merged for doing the work with code spaces right.

C

A

Noticed, in addition,.

C

Yeah his Fork moved right, like I, don't know where it went, but I was looking at it. The other day. Yeah and I wanted to finish reviewing it and it was gone. So I was like okay, apparently.

A

C

A

He was uh because to to build kubernetes, there's a lot of other dependencies, so he was working on sorting that out or potentially using like we have. We build an image anyway for testing that has all those dependses and possibly just like basically pulling that in as like a base image, so um I I think you know it's. It's live it's considered like a soft lunch, but I would really you know tweet about it. Do whatever once we actually have the code spaces component in there, because then that is a full. You know.

A

Here's your yeah.

C

Here's the interesting.

A

Tutorials exactly.

C

Window to terminal to type and off you go yeah. No, that makes a ton of sense.

B

Do we have a vague idea when that might happen.

A

uh The broad goal is to have everything done before: keep gone: okay, so April.

C

A

It'll happen sooner but like yeah, the the goal is just have all of this merged before kubecon.

B

But not like the pr is ready, it'll be merged next week or something yeah, okay, cool.

B

Thank you, sir.

A

Gotta find the tab again. There we go.

A

Okay, uh does anyone else have any other comments for docs.

D

Excited about that.

C

D

A

Anything for moderation and media, so YouTube zoom and all that other fun stuff, no yeah I think we will once the zapier is up and going we'll.

C

Have a lot, because that's going to do many, many things for us um to just to let folks know Nigel's account is doing stuff on our behalf yeah. So once we get this I intend to like merge all that stuff together, um because I know I'm using stuff to send a buffer, and all this other stuff, like oh there's at least four jobs that we're gonna put in this account like right off the bat or three I should say.

A

Okay, um moving over does that well. Does anyone else have any questions comments on that.

A

Okay, uh GitHub management, uh don't really have any updates here.

C

A

C

Need people like what's going on there.

A

It's like so getting the rotation everything set up was uh one of the tasks for the uh lead, Shadows, so hoping um something of that might come soon. Okay, cool.

C

Yeah I, when I get back I intend to be more proactive on the org repo for memberships. So.

C

I'll figure out how to do them all at once or all an individual commits and one big PR kind of deal.

A

Yeah we we have the little tool that does it yeah.

C

I've been using K org like on individual people, so.

A

You basically just do. Instead of you, create a branch for like ad members date, and then you add the member, then just add the next member go.

C

Through it on that PR nice yep, that's out, I.

A

D

A

The tool like more.

C

Than you could ever possibly imagine.

A

That there will hopefully be some more stuff coming to that uh soon. That will help actually like team management and things like that. Oh beautiful nice.

A

um So that takes us to open mic discussion. I do have uh one item, um but I will hunt that um and in case anyone else says anything they want to talk about.

C

A

uh Let me just give me one sec to get the link.

A

um So there will be some more comms about this going out uh relatively soon, um but we are look up guessing.

D

I was gonna, say: did you was this Open Mic discussion for the recording, or was this for after.

A

A meeting uh this this one is for recording cool uh the.

A

To be like just upfront, um we are on a trajectory to um hit 4 million in credit usage, um even with the changes we're starting to put in, um which is a a large problem for us.

C

Gcp you mean.

A

Yes, yeah, okay uh and it's it's a like. Even though we've talked about the registry, we're trying to boost comms about the registry, people aren't switching so.

C

Yeah um blog post that we already.

A

Knew we have, we have video, we have two blog posts, that's right! Oh man, all right and and I've already engaged the cncf end user, community um and I have um I'm, also working with Taylor and he's going to put together a tutorial on how you mirror the registry. So if you need to like, if you are behind, um you can at least have a local copy of those images um yeah which honestly people should be doing anyway. If they're running the stuff in production.

C

Yeah and you would think the resiliency that registry that case.io would offer people would be incentive enough, but I guess not.

A

B

I'm wondering if people have built their own mirrors and built their mirrors based on registry instead of somewhere else. No.

A

No, no, they have not built their mirrors. um That is as part of the problem um they are. We are, or you're still seeing massive traffic pulling from keats.gcrdio the thing that we want to go away.

A

So we're getting to that point.

D

It's probably the only thing: that's going to work honestly.

A

C

The thing is, we would be real.

A

If we didn't like television hold on yeah like so yes, um we are, we are looking to not necessarily turn it off, um but we are going to set a policy on Aging out old content good. um So right now we literally have everything from the beginning of time. Right like you can go back and people are still today pulling images of kubernetes from before it was even a cncf project.

A

So this change will break people.

C

B

C

Anything in the oci spec to allow for like message.

A

No or anything, no like there's nothing, we can do because they are running on Ancient stuff that still points to the new thing. There's no way for us to push out a change to them.

C

No I get it I, get it it's just. They literally need to just change this and I'm sure. There's hundreds of people not realizing that they're using it.

A

The yeah, the honestly, the the the the biggest problem um is the fact so many there are thousands and thousands of tutorials out there that are all pointing and using old content, um and the other thing is like people have hard-coded CI and things like that.

C

Or the 301 redirect break CI I, don't think so we well, it would change yeah. We.

A

Have issues there yeah? um We we can't do a redirect on Keystone gcr.io.

B

Can we do it as just a dnsc name.

A

No, we have we have. We have spent many hours like honestly uh we're talking sweet weeks of time, wow looking into this Jesus, so trust me all the things that you can think of. We've already gone down those paths; okay, fair enough.

A

Regarding the policy uh and how we are aging out content, uh the discussion is happening in that issue, I just linked. um That will be. If you have specific thoughts that will be the place where to direct them.

A

um We are. We don't want to surfaces widely to end users, because, to be frank like that, you know them just wanting to come in and hold content forever is not a sustainable solution. No um and.

C

It's our Harbor and all these other projects, our.

A

Our in addition to it being a problem in terms of um egress traffic and people pulling these old images, it is actually causing problems for our building release process because because it's having to walk through and tag things like there are um 90 000 images or something like that in there right now,.

C

A

um Oh and we we actually exhausted um I, was about to say um our we got like API throttled the other day, even even trying to do it so wow. It is not something we can like. We can potentially work around some of those issues, but again it's not sustainable and we shouldn't.

A

We shouldn't have to keep everything from the beginning of time, um but when we do determine this a if you have thoughts on this policy and how long to potentially retain them, which I mean on the issue, but the big thing will be when we do actually decide this policy.

A

um That's when everyone needs to scream it from every Mountaintop. They can.

A

We're looking at other options, um other things we can potentially do to lessen it, but this is a honestly an inevitability.

B

Anyway, everything.

C

B

Still exists, I mean it's still GitHub. The images aren't built but everything's still there. It's.

A

It's actually more difficult than you think, to reconstruct an old image um because all like, when you rebuild a container image now all the dependencies. So if you're pulling like you know from Debian.

B

A

Whatever all the base dependencies have changed, yeah, hmm like you know, we we at this point, survived the python two to three change.

A

I'm just tossing that as an example wow so like.

B

Well, I mean if, if somebody still really wants to run kubernetes 1.2, then they'd better have an appropriate build environment. For that you know, yep.

A

And the the biggest thing is that we will just need to communicate it.

C

Yeah, like a massive comms push I'm, seeing like yeah and even outreaching to some influencers that we know kind of deal right like.

A

Yeah, like once, we, the the uh once the cncf ambassador program, completes rebooting. uh That's gonna be one of their big tasks, cool um and the like. If we decide the policy before kubecon, it will probably get mentioned on stage good.

D

um So when this does get decided- and we start shouting it from mountaintops, the solution for folks who are pulling these old images who will be broken by this will be to create their own mirrors of that image, probably right yeah.

D

So if we are sending out a whole bunch of com saying this is a big change, that's going to break people and your solution. If you need this is to create your own mirror, we should we're suddenly going to have a bajillion people trying to create mirrors. Aren't we yep.

A

That's why, like Taylor uh Taylor, was going to write up a tutorial on how to do a mirror for this stuff.

D

Well, I'm just thinking that'll blow through a lot of budget. Won't it because a lot of people a.

C

Lot of people will be yanking now at.

D

The same time, but.

C

They might already be doing it anyway. Well.

A

They're not using mirrors at all yeah. um The other thing like you know you might have seen the blog post. We are not pushing any updates to the old registry, so all new images are only available at the new endpoint beautiful um as.

C

It should be the.

A

C

All the providers that are still running old version.

A

Yes, now at this point, yep um and the vast majority of them are running on top of AWS, like we know that from from the traffic, oh.

C

Yeah yeah I, don't doubt that at all yeah.

A

um The I was gonna say. um The other thing that we are thinking of doing is setting like basically having a Archive of all the tarballs of all the all of all the current images. So if you desperately do have to reconstruct something right, you.

C

You at least have.

A

C

In Time snapshot sitting there waiting for you, yeah.

A

No on sort of like unthawing it and actually using those will be a pain in the ass, but as an intentionally a pain in the ass uh to you know it's basically, just like the you know: okay, you've really screwed yourself. This is how this is an option of how you can get out of it, um but yeah I I just wanted to bring that up here, um yeah as a potential big thing um that is going to happen, and this.

C

Is bigger than Docker shim? Oh.

A

Yes, like far bigger, oh yes, um and to um if you have thoughts, look at that issue.

C

I'll, look at it later or.

B

C

In a few and try and get some comments in.

A

I think at this point we all agree it's happening. It's just determine the window of time for us to retain it.

A

That's the the big debate, um so does anyone else have any uh comments about that or any other open items.

A

C

Chaslin, when we're both back, we probably should bring this up as when you create like a Content campaign. Kind of deal. That'll include Taylor's, tutorial that'll include many many tweets and reminders emails all that stuff right like we need a full spectrum attack like wider than we did for Docker shim.

D

Good opportunity to talk about sensitive Communications with the.

C

Team as well yeah.

D

A

Any documentation we have on that. The the other thing like in terms of bringing up the like, don't bring up the policy publicly until we actually have a you know. Well, you know publicly being the brighter broader Community keeping it in kubernetes is fine yeah, um but yeah.

A

Okay, I will go ahead and stop the recording.