Kubernetes Storage SIG, 3 Nov 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Kubernetes SIG Storage - Bi-Weekly Meeting 2022-11-03

Description

Kubernetes Storage Special-Interest-Group (SIG) Bi-Weekly Meeting - 03 November 2022

Meeting Notes/Agenda: -

Find out more about the Storage SIG here: https://github.com/kubernetes/community/tree/master/sig-storage

Moderator: Xing Yang (VMware)

A

Hello, everyone today is November 3rd 2022. This is the kubernetes touristic meeting today we're going to go over the 1.26 planning spreadsheet and our next deadline is coming. It's actually next Tuesday November, 8th, that's code, freeze and, and then November 10th. That's the deadline to submit a doc placeholder PR, all right. Let's go over this!

A

The first one is the delegate FS group 2 CSI driver instead of cubelet.

B

um Hi uh so I have a PR out to to graduate the feature and it's mostly waiting for the uh final release of CSI spec. Okay,.

A

And inside you're going to cut the release today right.

C

uh Yes, that's the.

A

Okay, thank you.

A

um Right, it's this one and also the the dark PR is due soon as well. Can you also submit that.

D

One sure sure, okay.

A

Thank you, and next one recovering from research, radio, ETV test command there. Any update on tests.

C

Not not today, but I'll, try to work or find them to work on it. Okay,.

A

A

Next issues related to sim Rollins amount of points. This is the bug fix is a gene here, looks like she's, not here.

A

A

Oh yes, that looks like yeah. So has the audio issue just here half of what you said.

A

Hey uh next one CSI Admiral volumes. This is for the conformance test right.

E

Yeah right so the pr that I had opened to fix, with, like you test that merged and I, have one PR open to promote these to conformance tests, I'm just waiting on the two-week soap time. But that's ready for review foreign.

A

You would wait for one more week so, but you need to meet the there. Is this a test? Freeze right, you need to do you need to get merged before that deadline.

E

uh Yes and we should meet the two-week window by next Monday, so.

A

I think we can merge.

E

Anytime, after that, okay.

A

Yeah, okay, maybe actually you are actually not really strictly.

A

um We should buy the Sterling this, because the feature is already it's already there right. This is just to the that in the test.

E

F

A

um One week two week, one week.

E

One week yeah all.

A

Right thanks and uh next one, okay. Next one is done more in group, API yeah. So this one, um the there's still some questions. Some some review comments on the on the CSI spec PR I need to schedule meeting with the gyms to address his concerns.

A

Next, one is provision volumes from Cross, namespace snapshot, I think we are only working on the snapshot. Part.

G

Okay, API destination continue to discuss on GitHub and external production or open the issue and submitted Politics. The extra promotional code depends on APF specification.

A

So yeah I see you have this uh VR. This is still being still on resolved issues. Okay,.

A

Since Michelle has been reviewing us.

A

A

I thought I chatted with Michelle yesterday uh it sounds like there's no major concerns, but basically okay I think she said you're you're implementing it right. So probably that's that's. What's left uh is it the I know you open the issue and it's not provisioner. Is that the one that you need to implement or do you do you need to do more implantation entry or auditory.

G

Yeah, hello, oh okay, hi! Yes,.

A

I'm, just like uh the so so you need to make some more code changes in Trio auditory.

A

Okay, we can follow up offline, I'll, just say it's been happening. It's fine I'm, seeing reviewed.

A

um And the next one, the sporting house, Edition, Matrix and so yeah, so we didn't merge the CSS back PR uh so uh and then for the E2 test still have not heard of our rent. Yet so I need to Ping him again.

A

Next one is the new uh this right. This is um RW access mode, ETV test this one. We do. We I think we have some tests submitted, eraser or emerged yeah.

D

I think the test is merged. Okay,.

D

A

Thanks next one is a runtime, assisted, Mountain.

H

So far, uh this one I submitted the CSI PR, uh so I would love some reviews on that, and also following up with the signaled about the runtime class suggestions.

D

H

Came out of the last round of reviews.

H

And I think the target status for 1.26 should continue to say design because it's not going to be Alpha.

A

Oh okay, yes, all right so to design thanks thanks next one CS proxy for Windows transitioned privileged containers is Mauricio on the call or if you want a given date, yeah.

H

So I think uh Alex is making a lot of progress on making the proxy, basically just the library, so no more binary and a separate Branch will continue to release the binary. That's existing with, like any security fixes or any bug fixes if needed, is.

A

This the change will be in this existence, is a proxy report, or is it someplace.

H

Else, let's say the existing CSI peroxide, okay.

I

Yeah and I can also add that he's um has made a change to use the new library in the PD CSI driver. So, okay.

A

I

A plus one on progress being made.

A

Okay, great thanks thanks.

A

This is approximation from this issue, not sure if anyone is following up with this.

A

Yeah, so so we are basically doing some internal testing just to see what is the? What is the scale? What is what is the performance that we can support?

A

uh No extension, Secret is the humble here, um so I have not seen any oh I. Think last time we're saying maybe someone from Michelle's team is going to welcome E2 tests with some new contributors, so I'm not sure about this one I'm going to check this out.

A

There goes Michelle I'm, not sure about.

F

This yeah um yeah they're I, think they're. They were interested but I'm, not sure how much time they're good they're gonna have to get this done by 126 code. Freeze.

A

C

Oh, this is just we had. This is not targeting 126 anyway right. It.

A

Didn't it is uh this one it is uh actually it is a targeting.

A

Oh no, no yeah, you're right! Because, okay, let me make this one change this one.

F

A

To be there first right, that's right, yeah,.

F

So I think I think the main challenge is we. We would like to have to test in the 126 time frame so that we can try to promote it in 127., yeah.

C

A

A

I think he uh probably won't because I did check him, because he has a few things. Okay, it's targeting, so he basically says he he now he come back where he started to get back to those work, um but probably won't be able to not necessarily he can finish all of those uh on time. So I you know, did this regarding this test yourself, I did checking check with him last time he's satisfying. If somebody else wanted help with testing uh is the E2 test.

A

He has not started this one, so, okay, so this one's still, it's two to do.

A

And next one is a secure, Linux relabeling using Mount options.

D

So I have volume reconstruction PR on review so far, I got feedback from hemant and I'm fixing his comments and uh adding it to uh human test and so on. But I will also need some subsequent PR, so that will actually add a series to the Reconstruction. So I would like that a reconstruction PR reviewed like as soon as possible 16pr should be simple or.

A

Rather so you want this one to you want this one to be merge it first before.

D

Yeah another piano I can stuff everything into a single PR, but that would be even bigger.

A

Okay, so it's a chain would be good for chin. To take a look right. So I know you pinned her last time.

A

I'm not seeing she's not here.

D

C

I mean I think it's okay, if it is in the same period also, but yeah.

A

Or maybe you can, can you just submit another Pia but kind of on uh just to based on this.

G

A

It's a little Library a separate commit right. This way people can just review the new changes. I will submit.

D

A

F

A

Course this one is done, and next one is CS migration. This year um we do have a PR out, uh I. Think there's some question about the window, see the Windows support. uh So we did add a note there saying that it's playing into GA CSA windows in the next next release, see me and toga.

C

As the next release of CSI driver right right.

A

That's the one that we are going to validate 106.

C

So he's saying that, like we could uh as a team, the migration PR because we're gonna GA the CSI driver window support in the same time frame anyways.

A

Right, that's the one that we attest because see we we always test the one after the release right, so the one that we released recently is for 1.25. That's the next one will be for 1.26.

A

That is for for this right. So when we add this change there, we will be moving Windows to J there.

C

A

You guys have more concerns. If you guys are okay, I will go ahead and merge if you guys still have concerns and I'll leave it in for more questions.

D

uh Maybe a stupid question, so we will release kubernetes, but we will not have a CSA driver that is uh working with that kubernetes right. We.

A

That's that's the that's. The RC's driver is always doing the qualification after kubernetes release right. So if you look at the the car driver, it's not going to work for 1.26, it's not even released yet I can't really say it works for that. So we always test that after the the kubernetes released so like the one we just released, I think just like a week ago.

A

uh That's what that's added 1.25 support right, because we qualified with 1.25 not necessarily does not work. It's just so we have not qualified that one before that. So the next one we're qualifying.

D

But that's that mean that needs to be in release, notes at least like to warn people, but.

A

That may be strange right if you, if you started to add the uh like CSI driver release, notes into the kubernetes release, notes, that's also weird right. um So, okay, how do you? How do you want to phrase this I mean right now.

D

I, don't know if just people to wait for the for the season driver release before upgrading to 120 Scripts.

A

But then they, if they use C HR, they will be look at CSR driver release, notes they are not going to check kubernetes release notes for that.

D

Right but they are going to update kubernetes to 126.

A

They have to look at our scissor jar and see which one supports that right. So that's that's the next release. That's what we we add in our driver.

D

um That's strange.

A

Yeah but the QE went where not to sign off if uh it will.

B

D

So I have a kubernetes cluster I will just look at the kubernetes resistance. Okay, I can see everything is fine, so let's upgrade it and then I realized that I have something that's probably broken.

D

Well, it's all broken because we tested and it looks working but uh yes well.

A

It's it's async, that's the thing! That's! This is all it's all the tree. That's that you always have this problem.

C

A

C

I mean it is auto-ship, because the CSI API is supposed to be stable, I think the I think the yeah. The concern is like over the around the Windows support like if they upgrade and migration is on by default and uh and with the whatever version they are running CSR driver. Then they will be broken, so there should be something in if you're going to allow. Maybe we can talk this offline and and how upgrade will work yeah.

A

Maybe you can, can you just add your message on that? uh Pr will be easier. This way we can just sort out what you want to do over there. You need a release node or something or some fashion, that maybe add comments there, yeah. Okay, all right! Thank you.

A

uh Okay, next one remove entry, Azure disk, uh so I pinned Andy I, think he's on vacation, so I have not seen a PR. Yet is anyone seen a PR removing the entry disc.

A

Let me pyramid.

A

So I'm not sure if we got time but I did pee him earlier, he said he was submitted, but I'm not seeing it yet. Okay uh and then next one is the CSM aggression Azure file. So uh this one, the rest of this one we we do have a PR I, think we're just waiting for the the delegate of his groups in to emerge right. So yes, PR, pending.

F

A

And in the next two, the ceph, RBD and CFS I have not seen any update from humble I think he did tell me that he said he just got back may not be able to finish those so I'll just saying make no update. Unless he said there was a PR, but it's probably unlikely now for it's already getting pretty close to the code. Freeze.

A

Remove entry glassware first yeah, so this one uh this was removed a while back I just feel humble for the dog changes. He has a dog PR about PR out.

A

um It's actually, there are quite a few places that we need to update.

A

And next one is always on a reclaim policy. This is for E2 tests and some code changes is debug on call.

A

Okay, uh yeah. You want to give an updates.

D

For merged, we are just missing end-to-end tests.

A

A

Thank you, and next one is a control volume, mode conversion between source and Target.

A

um It's ronaka on call, so ronak has a PR on the release tools, repo, let's add this framework, so that we could add into your test auditory that one is merged to repeat.

F

A

He's working on working on tests on the provisional repo for just for this feature.

A

Next one is the numbers shut up um yeah, so so, do you have a PR out, promote future gate and also an LPI to add a matrix, so so uh has already reviewed it, and so you are fine with that right, so I need I. Think I need to find another uh someone who has the permission to emerge.

A

It's in the controller that folder to merge that which one the metrics one I, don't think that none of us are metric. We are already merged. Oh you sure, I checked this morning was not okay, okay, that's good, uh okay and then the future cater one. I think what we want to ask a gene to take a look, so I'll ping, her yeah.

C

A

A

Next one is enable username space in cubelet. Do we get any update on this? There's no update hold on. Do we know, what's going on with this sounds like no update for a while.

C

Yeah hi I haven't been.

A

A

Let's see no update, and next one is a PVC created by stapleset, will not be Auto removed. uh Net yeah.

I

We can continue in the march to uh beta uh so right now, working on the prr metrics are going to be done through Cube State uh metrics, but I just ran into a complication that it's not clear how they work with uh fields that are behind a feature gate so um and working that out. If anyone has any advice on using Cube State metrics I would love to talk to you over a slack or whatever.

A

F

Mean I think I think the cube State metrics doesn't need to worry about the future gate.

I

Yes, it seems that the end-to-end tests are failing because of the feature gate not being on.

A

I

A

Using this, uh you I think that you have this. um Gk has like Alpha cluster or some Alpha cluster. Are you using that one or.

I

No, this is so that the cube State metrics has end-to-end tests that run on on uh kind and like they seem to be pinned to like 124 or 125, or something so.

A

I don't know if you have a way to enable that then.

I

Yes, so I I have a discussion open with someone who may know, but um you know it's always good to have uh some more expertise if anyone out there here, okay.

A

I probably won't check unless you figure out what I want to check with you, because we're also looking to maybe add something there for the uh looking to maybe add some more matches there as well. So, okay, maybe okay,.

I

Okay yeah, if I ever become useful there yeah okay be happy to help.

A

All right, but you're, saying the main, the main PR is merged. I mean PR to promote what about the main PR.

I

um No still need to promote the actual feature date change, um I.

A

um I wanted to get.

I

The metrics change in, um what's that.

A

The pr is not submitted yet.

I

For the feature, gate change, yes, that that is uh correct. It is not submitted.

A

Oh okay, later.

I

Yeah I was hoping to get uh the metrics done, because I thought it would be quick and then it would be easier to submit the feature gate but yeah. Okay,.

A

H

A

A

And then next one is foreign extension for staple sets um to be sohammad. Do we have any update on this one.

C

uh No not today, okay,.

A

And then we have this one: better default story: class young.

D

So the code has been merged, okay, which will be this data, and we also I, wouldn't be it to write a block.

A

A

uh Oh also, this one: do you also need to submit a I think you also need to submit doc PR IDE placeholder PR by 10.

F

F

A

All right, okay, I think that's all we have. uh Let's see, do we have any? Oh okay, so Matt do you want to talk about this.

I

Yeah, so this was just this I, don't necessarily need to open a discussion here. I just wanted um to raise the profile and remind people that we have a discussion going on on this, the stalker room as well to contribute and that uh after the code code, freeze, I figured would make um a meeting so that we could just discuss things in person. But I think it would be great to get some more uh um opinions on the doc before having a face-to-face.

A

Yeah yeah, thank you yeah, so yeah. This is the following: the discussion in the in previous meeting uh on the modern modification proposal. So now we decided to narrow it down just to limit this to the qis first. So this is the proposal um QRS yeah. So if.

G

A

Yeah, if you're interested, please review this and provide your feedback. Okay,.

I

Cool. Thank you.

A

Thanks anything else.

A

If not, then that's it for today. Thank you, bye.

I