Kubernetes Kubernetes AWS Provider Subproject, 18 Oct 2019

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: kubernetes sig-aws 20191018

Description

kubernetes sig-aws meeting 20191018

A

Hello, everybody today is Friday October 18th. This is the provider ingress meeting I. Am your moderator facilitator, just in Santa, Barbara I work at Google reminder this meeting is being recorded, will be put on the internet on after after the meeting ends and completely mindful of our code of conduct, which boils down to be a respectful and good person.

A

The we do have I think one or two items on the agenda, which Nick is adding as we speak, but please do feel free if you have other items to add them to the agenda and do clean your name on the list. If you would like to annecy powerful for anyone watching the video afterwards and otherwise, I will in any case, I will hand the floor over to Nick and, please do add, other items.

B

Cool so the first one I put a specific issue there, but before it's talking about that, I just wanted to say that so I think we should sort of keep this SIG alive because I have been, or whatever it's called provider working group whatever, because I haven't it's an indistinct up above provider meetings and we don't really get into the sig specific or provider specific stuff as much. There is a short time for updates where we can say what we've been looking at.

B

What issues we've been you know dealing with, but definitely don't get into the nitty-gritty of the cloud providers specific stuff. So personally, I haven't been keeping on top of AWS issues. I was just looking through a couple, and so I was just. This is a really easy one that I figured. We could just review right now, but I did want to have that kind of bigger conversation of how do we stay on top of it'll us issues and I?

B

Think it's really just each of us should take some time every week and make sure we're you know pinging each other. If we see important issues and and just you know, staying on top of it any thoughts on that anybody comments.

A

I read indicated know if you're raising your hand there Kendall, but yes, I, agree. Okay, I agree entirely. Are the I think what we decided is that we will do what other SIG's do where they effectively? If nothing's on the agenda, they'll they'll put out a call for like last call for agenda items and if nothing goes on, then they do cancel the meeting like with 24 hours or 48 hours.

A

Notice system like that I was remiss and didn't do it this week, because we did that last week so and I think it's a good thing. Cuz some people. There are some things we should discuss and but yeah. So if we do that and then I think out of that will naturally fall a more natural cadence, and so it might be that we discover that, like every four weeks is actually a natural cadence that works well, but we can.

A

We can at least for the next couple of meetings or whatever appropriate time period is we can use that discovery mechanism to figure out what that right gains.

B

Back cool yeah we should probably get through cube con and then maybe early next year we can figure out what the kittens should.

A

Be maybe yes, almost 2020, that's a scary thought yeah.

A

B

Yeah so I don't know if you got a chance to look at it yet, but it basically boils down to selecting the first, only the first port for HTTP health checks and that's not necessarily the correct one, and so this person solution is to just iterate through them.

A

B

A

Take a look when I'm, not under pressure yeah.

B

Five seconds.

A

Exactly just because like getting it wrong is is bad so that I yes, this looks this looks correct and it looks reasonable, I think just looking I'm, not sure whether like it looks like oh because we always use SSL I was just noticing like HTTPS and SSL, we're sort of like either one is acceptable for either one so I, don't know whether that's okay in general, like anyway, we.

B

Can anymore, okay.

A

Yeah, thank you bringing it up, yes, and if there are other issues than yes definitely we should. We should forget a way to better triage them. I feel like in general, like triage of issues and pull requests in the Khmers repo is very, very challenging. Yes,.

B

I totally agree: it's really easy to it's. It's kind of like either you're getting that you have a good email, filter and you're getting them through there or you're.

B

You know like going and looking every few days or something like that: checking notifications and you're on the Segovia's miss whatever, but yeah I think I I would like it if we came up with some sort of process, but I don't think we have to do that right now.

A

Another item about the out of cloud provider yeah.

B

This is just so like last week or something we merged a pair of mine, which is just to go back to importing the legacy cloud provider, so I think sort of off the top of my head and and what people have been sort of the overall cloud providers. Sake has been talking about in terms of timeline. It seems like it's kind of like at least for the the mechanisms in kubernetes to allow for migrations to external cloud providers.

B

It's alpha, 117, beta, 118, ga 119 and then ripping the code out in like 121, or something like that. So it's just something to kind of keep in mind. I think we like upset AWS, are definitely going to start pushing that forward. So at some point we will need to come up with a time where we migrate code out of the legacy cloud provider into the other one and.

A

To be clear, the it used to be that the kubernetes cloud provider, AWS external out of three project, had copied yeah the code and now it's simply reversed.

B

That renders the go okay.

A

That's great because now we we now when we go to make that change, we know that the code was the same. Rendering is like yeah, it's the same, so we only have to change one thing at a time: yes, except change where those controllers run. We don't like that makes a ton of sense. Thank you, yeah! That's that's great! That makes like yeah, easier and I guess: I guess we should figure out. I, guess I understand, like the big stakeholders would be like getting coo Vidya and going getting cops going. Getting yes going.

A

That will be. Those are the big thing.

B

A

Then but yeah contest coverage, yeah.

B

Test coverage is, is the big I think next step I think that's we didn't I mean we can make up for a lack of test coverage on the entry provider with the auditory provider since right now, it's the same code.

A

That's great, but yeah I think do we know if she is planning, do you know? Is it public if GC a third? When do you see I can predict me? I can track it down but, like we know, GC ease plan for removing it's.

B

Like everything that I have gotten is very similar to what Walter and it's basically Walter what Walter is planning so I'm sort of following his timeline and so yeah I think they'll be the same. Okay.

A

And then yes, I noticed at the bottom I think there's CSI is related. The CSI extraction is related, but not oh, it's somatically related, but not necessarily bound in any way to this I believe yeah, okay, perfect yeah. Actually that's great news. Thank you for that. That makes you so much better that it is that yeah. That's a good step.

B

Cool, that's all I had oh.

A

Thank you there's nothing else on the agenda. I don't know if anyone else would like to add anything. Otherwise we get a lot of time back. That's very, very productive, yeah, I.

B

Mean they anyone else is nice. Oh no, I'm, good, I might.

A

Come up with something you never know, I often refer them to open the like anyone else. Third party, oh yes, Justin.

B

Justin recognizes Justin Justin.

A

I give a yes. If not, then I will stop the recording in wish everyone a very happy rest of their Friday and every weekend. Thank you.