Kubernetes Storage SIG CSI Implementation Workgroup, 5 Oct 2018

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Kubernetes SIG Storage - CSI Implementation Workgroup 20181005

Description

Meeting of Kubernetes Storage Special-Interest-Group (SIG) Workgroup for Container-Storage-Interface (CSI) Implementation - 05 October 2018

Meeting Notes/Agenda: https://docs.google.com/document/d/1-WmRYvqw1FREcD1jmZAOjC0jX6Gop8FMOzsdqXHFoT4/edit#heading=h.csr1vkipn8s

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

Moderator: Saad Ali (Google)

A

All right today is Friday October 2018. This is the meeting of the kubernetes es on implementation group. As a reminder, this meeting is public record and published on YouTube.

A

Today, we're gonna go over the remaining tasks that we have outstanding for 1.12 and then do a quick overview of the tasks that we ask for key for this quarter for the CSI implementation.

A

The last items that we have left on the 112 112 release are getting the 0.40 sidecar containers cut in tagged. There was a open issue that was blocking the release of the 0.40.

A

Sidecar containers but Yohn took a look at it and he introduced a patch in kubernetes kubernetes, and he explained that 0.40 does not need to block on that and that we should be able to go ahead and merge or I'll go ahead and cut a new release. So that was the signal that we got yesterday as if this point it doesn't look like release, the sidecar containers have been cut yet I'll follow up with Luis after this meeting to see if we can get that done today,.

B

So I updated it dependency for that the driver register right now, but that was um that was before the one dot child odd Oh tag was available, so it's actually pointing to the release branch.

B

So this one we smudged it do I need to go update should right now. Basically, it's pointed Orange try, but that Chum pointed out that orange can get changed. If there is something should we change that you this tag that this tab just to come comes out, the 1-0 Ted. Let's do that I'm, sorry, yeah! Let's! uh Let's do that? Okay, we don't need you I, remember you says something we need to watch out da one. We don't need to wait for that. One right, not.

A

Anymore, so I thought we needed to do that. We don't need one twelve one anymore. Let's just go ahead and point it at 112, zero tag, okay, I'm just think up with Louise to make sure that you get this in before he does the cut and then give them a thumbs up once all of these are completed. Okay,.

B

A

A

And you're gonna do it for all the sidecar containers, yeah.

B

Yeah so yes, I'm emerged and an external storage yeah our sink, always a chair again, because the the one chopped dog, zero tag was just become available. So you know updated that one okay.

A

That make sense.

A

And any updates on the end-to-end tests.

B

No not yet it's a chicken says she's going to look so I wallpaper, again sure.

A

B

A

Sounds good and then, let's move on to the current tasks looks like there's only three of us here, so we'll only be able to get the status for these three C or D installation mechanism. I followed up with API machinery. They told me to put it on their next cig meeting to discuss, which will be Wednesday next week. So I'll go there and discuss it with them.

A

It's possible that they'll for hurt me to say cluster lifecycle to go and talk to cluster lifecycle, because it's an installation, bootstrapping problem, not necessarily a API Machinery core problem. In the meantime, I think what we want to do is.

A

For the security team, what they ended up doing was using the add-on manager to install the CRD. So if we can do the same thing, that would unblock us for the short-term, at least, and that way whatever these other teams decide to do for the installation. Common installation mechanism were not dependent on it. So I'm going to pursue that note here.

A

And that's all the update I have on that deep. Do you have anything on technology support, yeah,.

C

So basically, Chang and I are working on it and we we identified a couple more work items there: okay, basically the first one is ongoing and we came up with basically Chiang file, an issue around potentially coming up with external admission webhook to validate some of the semantics of the topology T's. Basically, the map and set validations, everything's good, so I think it's probably something that I'll have to work in coordination with CR the installation process.

C

So most likely you perfect yeah well, probably have to sync up on that and how exactly to bring it in okay,.

A

C

I also would like to become a member of the external probationary repo.

A

C

A

Good ping me after this on slack and then all I'll add you.

C

That's all I had.

A

Cool and next up is snapshots, so is there any work in progress at the moment or so.

B

So I have just started looking at the finalizer.

C

B

Jing is looking at the reclaim policy: okay and the demission demission control that that's being reviewed basically just needs some updating, so the work is already done and sees the topology. So I was, I need a PR. Do the siesta soon perfect.

A

It's like good progress, Thank You, Jing Thanks next up is moving CSIR back role. Definition from kubernetes kubernetes to external repos I need to find someone to work on this I was trying to see if Jordan was going to be able to pick up this work. I haven't been able to get him to commit, so I might need to find someone else, so this is still in progress.

A

If any of you are aware of anybody that wants to work on this, let me know otherwise: I'll continue to look CSI entry moving the CSI driver and CSI noted foe object from alpha to beta. This is something that Theon is working on, but he's blocked on the CR, D installation mechanism being in place and I think we also need to think about if the CRD mechanism is going to be alpha this quarter.

A

Whatever this new mechanism is, are we okay moving this to beta as the installation mechanism, for it is alpha, and so we'll talk about that once we have a better understanding of how we're going to proceed with the CRV installation.

A

Next up is the CSI pod inline stuff. So that's not here we'll leave that and we'll leave the stuff that you assigned to Lewis moving CSI to 1.0.

A

There was a meeting Wednesday morning to discuss the tasks. No real follow up there other than the CC la's are being signed at the moment. That's it splitting up the driver registrar's. We got to reassign this too.

A

A

I'm so hopefully we get update on more of these on Monday, most.

B

Different is this different from the one just a ballad like two lines about this: wake up.

A

Driving good touch.

A

Alright, that's one less item for us to work on. That's pretty much it anything! You guys want to talk about so.

C

David and I met up yesterday around the migration story that he's working on ya know the next steps. There are probably like he's going to work on a translation foundation that I guess the provision around that I stay detached logic and build on top of okay. All the existing things can yeah so easily like existing PPC provision through the through the CSI equivalent.

C

But if it's not there, it'll do the right things so, but I think like timeline wise, it's probably not something that can be wrapped up by 1:30, so most like it'll, probably have to go over a bit. Okay,.

A

C

A

Is okay, so I think the stretch goal is to get it alpha. This quarter, okay,.

C

A

Up slipping then we can push it into q1, ideally the faster that we can get it done the better. But let's see yeah.

C

So mmm sort of tracking two main drivers like one is obviously the gcpd and the other one is the EBS since there's a fair bit of traction on that.

A

Did some PD and the other one the.

C

Scsi driver from steam so that you know we can try it try out the the translation mechanism, actually that.

A

Make sense: yeah, okay, cool sounds good.

A

Alright, that's all that I have for this meeting anything else. That's.

C

A

Thank you both and we'll see you on Monday yeah.

C

B