Kubernetes Data Protection Working Group, 5 Oct 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Kubernetes Data Protection WG - Bi-Weekly Meeting 2022-10-05

Description

Kubernetes Data Protection WG - Bi-Weekly Meeting - 05 October 2022

Meeting Notes/Agenda: -

Find out more about the Storage SIG here: https://github.com/kubernetes/community/tree/master/wg-data-protection

Moderator: Xing Yang (VMware)

A

Hello, everyone today is October 5th 2022 this sorry.

A

This is the kubernetes data protection language meeting uh so yeah. Today we were doing some cap updates and before that, I just uh uh want to talk a little bit about cubecon, which is coming at the end of this month.

A

um So for anyone who is attending coconut, America, um Shantae and I will be there. We have this data protection working group session on Wednesday. There is also a sick story session on Friday and on Monday uh there is the contributors Summit. We also have a contributor meet and greet um Friday, so those are two con. Do.

B

We do we have a session on the contributors um Summit.

A

I did not submit one.

B

A

Because uh yeah, because I don't see, I, don't have a lot of people going there sure um and also I also have another other than this. One I also have another event to go to nearest data on kubernetes Day event right.

C

A

Plan to go there as well, so yeah do not submit anything um and I do have a specific topic that I would like to talk about. But it's not really a data protection related. So.

C

A

Part of no shutter uh that I do plan to talk about that for with a few people at a contributor, Summit, okay,.

C

A

If anybody is there and you're interested in talk about anything else, uh you know we can meet there because I think they do have some tables there. We can just grab a table and talk.

C

A

Right, we have to be like a formal session, yeah yeah.

A

Okay, all right uh now, cab updates, uh Yvonne. You want to talk about CPT.

B

Yeah, um the the update for today will be short, um so basically I just been um uh just uh chatting with um Jan from Red Hat about like um some feedbacks. Yes on the um yeah, the CBT, aggregator, API, server, um implementation and stuff, like that and all of those um are recorded inside the cap as a review on command sum.

B

So if anyone is interested feel free to take a look but otherwise um some feedback from me this morning, so I'm going to update them um shortly, so I think that's pretty much it otherwise, like um I'm still waiting for um uh the pr folks to um review the yeah, the prr section.

A

This release I mean for this.

B

Not yet for 1.26 for 1.25, like um yeah I, think like um uh John, already, um okay, I I, don't know if you will get to it, so maybe I can send.

A

A problem reminder right: normally, they would like to see. uh You know sick, look good to me before they can, but this they can still review it. I think yeah yeah.

B

Yeah at one point he told me he would get uh he will he'll put it on his list, um so I mean for what it's worth like um you know, I can just like update all the changes like um you and yarn provided first and then I'll helping him.

A

Yeah I look at it this morning it looks okay, it looks different from what I look at it. Last time I seen there was a new. This new thing got added I, don't think that that was there before the.

B

uh Yeah so driver.

A

Discovery thing.

B

No, it's not. We talk about that like um in the previous working group,.

A

B

We did talk if you do it just do a quick search on you know, driver Discovery, okay and then now it's yeah. So it's the same thing that we talk about like this is how the aggregated API server um find out where the CSI driver is, and it's at the end um inside. It is really just like um a service. It's like a service, client config. It's the same thing as the um the admission web hook, like you know, just give it like a service endpoint and a plot and a path.

B

Okay and then like that's how like um yeah, the aggregate API server will know where the CSI driver uh is and on the CSI driver said, there's a CBT sidecar. It will respond to the requests from the aggregated, API server.

C

B

And then like, um if you scroll up a little bit to the risk risk and um assumptions, some like all the way to the top, there um I just keep going like near the top of the page uh uh yeah right here, yeah so like um so this part is um It's relatively newish. It's just like um you know, I, don't necess know we if we need to implement all of these for Alpha.

B

Basically, um it's just ways to protect both the kubernetes API server and the aggregate API server, because potentially the response, payload that comes back from the backend storage provider can be huge right.

B

um We want to make sure like uh the educate. Api server doesn't become like a source of task. You know, like someone just write a script to to to like just keep hitting like on the back end, and you can request like a return like hundreds of megabytes of data right, then we're really bad, so we need a way to mitigate those kind of um whether malicious or unintendent, like um you know, dark attempt.

B

So the way we do I this is where, like I, think, potentially the the pr team would provide some feedback as well. So the way like um we do that with be uh to tap into, like you know, the API server uh flow control, like of the the AFP API fairness and priority life feature.

B

So we need to make sure, like you know, our educated APS server has some sort of rate limiting uh feature using the kubernetes AFP um thing to protect ourselves from dos uh DDOS. For that matter, and then, like um you know, and then the content, the connection, the latency between the kubernetes API server and our aggregated API server uh needs to be mitigated as well.

B

So um so, there's quite a um the uh so yeah, like um you know, like uh probably we'll hear some feedback from the pr team about this I think this is um important. Definitely put it down here now. Whether like is something that we need to do, something that we can postpone to Beta.

B

um You know that is open to discussion.

A

I also added the API reviewer label this morning, because Yahoo was asking uh someone from the API reviewer side to take a look, because this is really new yeah as soon as they complain you, but it's just you know we normally, we don't. We don't implement this way right. Just not, then many people have experiences with aggregated APS server, so good to have someone from the API Machinery side to take a look just to yeah.

B

Yeah I think I think that would be helpful, especially in regards to the risk of medications that we talk about here, because, as we talk about earlier, like essentially like kubernetes, API server would still be in the um the request path right. Everything was to go through like the kubernetes API server before it proxy to our aggregate API server. So we need to make sure um you know we don't um at, like um you know, like a source of contention or like just a point of weakness um along the path.

B

So thanks a lot thanks for adding that label, but otherwise yeah, like um this um I, think um we are ready for more external reviews from PR and API groups and now, of course, in in conjunction to this. We also have that um the CSI API PR oud.

A

Yeah, are there any more comments? After no.

B

I think I address like um stuff that um have been provided so fighting under the container storage interface. Okay should be near the top of the list. I can remember, but yeah like the second one right. There I think humble also provided some inputs there, but regarding to like the CSI Secret, but otherwise, like um I, think we, you know just again waiting for more inputs from other folks.

D

Okay, hi uh can I, ask a question sure hi, my name is lagu I'm from cloud Casa, a backup company uh I haven't attended. These calls for some time. I mean are not actively so I have a couple of questions, or maybe one question in this regard. I remember quite a while back. There was a discussion about CBT for file system, TVs and I. Think recently, two three times back when I attended. There was the discussion about the block PVS. So are we considering both both block and file system cbts? uh As of now.

B

For for um for the purpose of this particular cap, node, it's just block, but change files is definitely on the list. um I guess, if anything like the goal is to make sure whatever we do here um in some way, shape or form can be helpful and re-reused for change files at.

C

Least at the API level, so just to confirm so this is block storage if PBS can still be file system or block.

B

This one is block just block four volumes: yeah. Oh.

C

Just block volumes yeah.

A

We're not really talking about changing file list, I! Think for no.

C

No I meant block storage, but the PVS can still be file system PV straight.

B

Yes, yes, so yeah, it.

A

Has to be blocked, and you know the lower layers be blocked, but on top it can be either roadblock mode or fast system mode. It can be exposed to Containers as a file systems right.

C

A

Yeah yeah, okay, good point: I'll share that type yeah.

B

Not like a network file system, yeah.

A

For that we do need to I think that only supports change file- I, don't know I, don't believe that can we can use the change block for that right. So that's, but that's something I I think there are some discussions here uh about like making the names more generic, so it can be extended to support a file right. There's some yeah.

B

I think that part is um put specifically around the driver. This driver discover is the Rd I. Think that part is just more like um you know, just uh calling it like. Maybe, instead of calling it driver Discovery, we call it something like um service configuration or like um volume, Delta or volume snapshot, Delta service configuration so more in alignment with, like you know, things like validating weapon configuration retaining weapon configuration. You know just more consistent with like um I guess, like um the the naming convention upstreams.

A

I think I I've seen Yael has a suggestion calling it like.

B

Yeah, if you do not shot something um right.

A

Yeah right here, yeah.

B

A

So he says: okay, so driver Discovery, glucose volume, snapshot, doubter service configuration, yeah, I. Think that's.

C

Probably better.

A

Yep, um let's see good all right yeah, so this one, if we quad like something like this, then right: okay,.

B

Yeah then we can reuse it for.

A

uh It's covering the name is kind of weird yeah.

B

A

Is probably better um yeah unless, if we think this can be extended to do something more than just a snapshot, I mean that that's the only reason we only call this. You know we don't want to say snapshot here. We don't need to think about this right so right now, this is only for is that volume snapshot data, whether in the file or blog? But could we use this? Do we intend to use this right now? We do not have this use case button clearly.

D

A

This structure for something else, yeah, that's something we maybe need to think a little bit, I'm, not sure. Do we we right now we do not have a use case.

B

Yeah I I, agree: I think um you know I.

C

B

Yeah I've, been like um this yeah I, think like coming out with a good name is especially for this one. It's a bit hard, so I think we can just keep it like the way Iran's suggested for now. If there's interchange, then you know, there's always a chance to revisit. Hopefully,.

A

B

Yep I think that's basically it from the CBT on kept perspective. Any questions so far.

A

So this particular crd will be already somewhere in kubernetes. Csi I guess will be part of that. A big CLD will be there.

B

uh Yeah, it would be part of um CSI.

A

Well, CSI, that's the that's! That's those Japanese interfaces right, but the crd will be part of kubernetes CSI. I guess do.

D

You mean, do you mean, along with volume, snapshot theories.

A

Right, maybe I I, don't know if it should be there or if, if it either should be like a different, separate repo, maybe I don't know we can think about that later. But it's like should be somewhere under kubernetes CSI should be one of those sidecar repos or.

B

Maybe yeah I think like.

A

Even reported under that org I.

B

Would think like I'm just need to stays with um the aggregated API server source code, wherever wherever.

A

We're going to Delta I think that you have no no two one is in this one, and it was another one right, yeah, the volume's.

B

Not shut down so.

A

This one I also needs to be somewhere right. So there's a.

B

A

Other one, this one.

B

Yep, it's up there yep right yep. If you keep.

A

Scrolling right, so what is Snapshot Delta Rhino data I know it's a discovery, so two of those probably should be together somewhere yeah, either part of a snapshot, people or could be a separate one. What can we can talk about that later? Okay,.

B

I think if it is, if this is all on the like on the snapshot feature, if maybe you can go into the same repo as the external.

A

Right, the only the only thing is I was just thinking about. Like the snapshot you go in there. That's uh that fulfill line is already that's a ga, so basically this one, it's yeah, I think it might be easier actually to have a separate report for this one. It's just because the way uh the code is structured right now, it's getting confusing.

A

uh If you look at this.

C

One yeah yeah.

A

Because we viewed the client and the server separately.

A

So if you look at this one right now, there's.

B

A

We this actually is like a uh I.

C

Think it's understood this is.

A

Actually, like V6, so it's actually like a V6. So if we add in this one, maybe confusing you know how to you know um when we, when we uh let's say when we have this client package.

C

A

Maybe so confusing.

D

A

You see this one. This is a client. This is a V6 right, so this might be confusing. If we add it here so, but we can talk about that later. We don't have to get this one yep. It sounds good. It's not here. Yeah go ahead. You have some questions. Yeah.

D

You know just quick feedback I think in the long run it could be better if they're all together, because even today you know a snapshot. Crd is being outside kubernetes. There is always some confusion right because you have to install the theories yeah separately and we as a backup company. We frequently run into this problem because a lot of people don't properly install it and we issue the volume snapshot. Call, and uh you know it.

C

D

In problems it doesn't reconcile and it just times out and after sometimes we need to figure out that you know there's some issue with.

A

Explanation, yeah, but even in the crdc in store, you still have the problem with the snapshot controller right. But that's awesome right.

D

A

The architecture problem that we have a product yeah because they don't want to include this one as a core API. So this is the Rd I think that's all right! I, don't think this is going to change it's going to stay here. Crd, it's just uh yeah, I, completely understand yeah! Definitely that.

D

That was to be my another question, so this is still external right. The volume snapshot, even though the feature is GA. It doesn't mean that the crds are installed with the stock kubernetes. No.

A

It will not go in there yeah.

D

A

This is the trend right, basically CR, more crd, swap as long as it works, unless, if you can prove that, oh there's no way, this can work anyway, this works. It has to be injury. There.

D

You have it looks. It works yeah. The.

A

Cigar patch is not going to accept this one to move back right because the kubernetes is getting so big, so they want to move things out possible. They also want to move persistent volumes out.

D

A

Was exactly like when we a few years back when we first, you know, brought this up in seek architecture we were proposing to Addison entry right, that's exactly what they said. Someone was saying: well, let you move. What's the volume out.

C

A

That I think that ship has already said what I don't think that will change, but this one uh yeah I think since this, but we definitely should have a better way right. We probably should bring this up again just to to I.

D

Mean it'll, be nice they're, not good.

A

No, they are not going to move this back, but they should have like some other ways to think about how to how do we make sure you know all the distross include this it's basically, these children need to include this when.

D

They install right, I think at this point the CSA is so standard that you know expecting. People to you know to do a separate, install is not good, but anyway I'm not I. Don't think this will change I'm, okay with that, but what I was about to say is with the introduction of these new crds. We are talking about like volume, snapshot, Delta and things like that.

D

Having another repo in another place, you know, will only exacerbate the problem, so we might as well reduce the scope by combining everything into this repo, because they're all related right, at least.

A

They're in one place, I'm- actually thinking, maybe the well. We can definitely we don't have together this Resort, but I. Imagine thinking, maybe actually is good for us to start using a new repo, but maybe eventually move the snapshot CRTs out.

A

um So the controller everything still will be in snapshot, but the crds will be a separate report, something like that, but not sure.

C

A

Reason I'm saying that is because right now the the package like there are two gold mod we're actually having some other problems. Because of that, so you see here right. So we have a go mod here, so we build this client separately uh and then you know, and then there is a server package recipe, the controller and sidecars we build here. uh So there are already some issues because of that.

B

Yeah well like the um I guess, the the versioning of the software or the snapshotter is different from the version of the apis right. It's a house organized Community, right.

A

But I think right now they are kind of mainly in sync, but but it's actually, we actually uh um release them separately, because there was a. There was a client one. So if you want to use the client itself, there was actually a client.

B

uh Because the the client or the the software can still be 6.1.0, but at the API level we still say Alpha, One V1,.

A

But we did one most of the time we released them together, but some occasionally they're out of sync.

D

Like what exactly the client do, what exactly is the.

A

Client this client, basically it just- has all the crds or some people want to use the client themselves like the. If you look at what what is in the client right, if you look at this client.

A

They want to use the client apis directly, then they can just yeah. They can just have this one. They don't need to use the server side all right, so those are so those are separate right, um but of course the downside is then they're not in the same repo. Then you have to do three separate release or you.

A

You know you you build this here and then, when you test it, you have to go to separate repo, that's also a pain yeah, so the pros and cons of each but right now, I do see some problems when I was you know releasing this one. uh Definitely I see some issues with this like even when we were running the release, notes tour.

A

It got confused, so sometimes it does not generate the results that they were, that that we need, because of this confusion, but that's that that we don't have to you know, get that decided here, but that's just a thought.

B

Yeah sure, but that's fine yeah. We can because.

A

Otherwise, if I- if we put that here, then this is already V6 right. So then what what is then? What do we call this? One I just thinks it's getting getting confusing right so like if we, when we're doing this one right. So we call this. You see this client package. We call this one V6 now now with those new apis that we start with V6, it's kind of weird yeah, I.

B

Think it's, it might be. Okay, because V6 again is the the version of the control, the the repo.

A

Okay, it's a client, it's the version of the client, because.

B

If you go into the client folder like even like the client themselves, they have their V1 Alpha One API level um right.

A

B

I see like volume snapshot one V1 right at the top there, so it'll be volume, slash a Delta, V1 Alpha One. If we edit here so.

D

B

At the API and the client level we are still a version separately, but you know we yeah I yeah.

A

Okay, we can yeah, let's get back.

B

A

B

Okay sounds good um cool.

A

um I think uh well, we okay we'll go back to here um so uh foreign file.

A

Change it files no.

D

No blog is also good, I mean that's, that's more efficient, definitely yeah, but the reason I asked is uh sometime back. There was discussion about that files. I.

E

A

Want to clarify.

D

If you're talking about the same thing or something.

A

Different yeah- we can get back to this this one after because, right now, this cap is already very, very complicated, yeah.

B

The last do we have any like examples of um change file Solutions out there I think. The last time we talked.

A

About oh yeah yeah, it does not look like. We have enough examples of that. Yeah.

B

And I know like dale dale has like um one FS or something.

C

F

B

Like the HP I, don't know like I, don't know, I don't think AWS has.

A

um Yeah, if anyone here, if you, if you are you, know, know some storage system that supports that. We can also look at that so because we want to design something based on some real examples. Yeah.

D

Yeah, currently, we use something like Block Level diff, which is okay, I haven't even heard about it, but that's that goes.

C

A

The blocks too,.

D

Yeah we are using copier for the blog.

A

But I thought coffee does not really give you. Those uh doesn't have an API to give you those blocks. Does it.

D

I mean it actually transfers to S3 storage, but you can you can intercept the rights and you can you can get the list of blogs if you want, um but we back.

A

Up here supports change your files, I thought.

D

No, it doesn't even in the files it'll actually compare. So it's a block level dedup, but you still has to read the file. So it's not as efficient as having the CBT from the storage itself. Okay,.

B

I think, like a call from Caston, um you know about it to your car. Like um does copia, do like chain file or is it block level.

F

uh Copier does file based differences; it three walks the file system and does a stat on every file and all compares the sizes and modification times.

A

Okay, but it doesn't really have an API to give you those right, I.

F

Don't know it doesn't have any pair to do it, but copia is open source and you can uh I mean the algorithm is straightforward. It's a tree, walk right and.

A

Yeah, but you can basically you can you can change it.

C

A

I'm just saying it does not have that itself, that's what I mean don't have API to retrieve.

F

This right, but that is like automatic requests. You can always request copia to open source. You know to for provide more formal apis over things. This part of the of copia, the I, think the tree Walker part is probably not too difficult to pull apart. Okay,.

B

Okay, so I guess like shin, sounds like there is something out there, but you know it's not like.

F

But the most important part about that is it's diff against. What, where is the previous state kept I mean that's the most important part you need the previous state.

D

Right, so that's why it's eucopia is back up to F3, so that's what it Compares. So it has to do that and obviously, if you have the diff from the stories, which is, that is far more efficient, yes than reading the file in comparison to the content right.

F

Because you have to pull in pretty store previous state and pull it back, so you can compare so yeah. Oh.

D

Yeah, you don't get the whole whole.

F

Data but you compare it yeah, it's it's some signature of whatever yeah.

B

So and and copia works with our file shares and network file system as well right, so I guess sit down a layer on top of the actual storage um yeah, true, okay, yeah, okay yeah, so so sharing sounds like I'm from a use case and user scenario's perspective like um to support change files. They are sounds like there are some use cases out there, um but maybe like.

A

Right, I, don't know someone we like to see the like. What are the apis, because we want to see you know if we design something it can be used. It's not like yeah. We know that copia probably has this use case, but we also want to see like how does the API look like right? That's what we can see from the power skill is one one example we can look at, but.

B

A

Have more examples, just just to see how the API looked like yeah.

B

So far the only concrete API we have was the one that function a while ago from Dale um I want I, don't know if um I can't remember, maybe EFS AWS EMS might have it, but I haven't looked too deep into it. So but anyways yeah. We can do some research here.

A

The one that you uh you you tested with with the you know the. What is that the uh the EPS.

C

A

All right, okay, yeah.

C

A

Okay, anything else.

D

I have one question but may not be directly related to this cap. Sorry, discussion about this cup is done. Okay again, this is going back to small discussion. I remember there was a cap about. You know, execute execution inside a pod. Some kind of standardization anything happened with that.

A

No that's stale, because uh um there are some concerns from both sick node and the API Machinery side, uh so Shan, China and I. We have not really got a like.

A

Basically, when you come up with some reason and then go going, go back to them and then you know try to persuade them to just to not get to that point. It's a there are some concerns. So right now it's stuck so that's the thing too.

D

Okay, yeah again.

A

D

Asking from the point of view of, uh if you want to freeze the application right, please.

A

Yeah yeah so right now you just have to do it, but you probably just to use the.

A

If we can move forward with that, one um yeah right.

D

That's how we do we do it, but.

A

I, just okay, yeah yeah. Unfortunately, it's um uh yeah, it's mainly I, think it's mainly signaled side, yeah concerned. uh You know if we add some big change like this, to signaled we're going to make their code base uh unstable or something right so just have to have a.

C

A

To say hey, so why not cause problem, but of course, if you add new code, I call always a couple questions of it. So that's that's, and also there are some apis that concerns yeah. There is a.

D

Security right, yeah, I, remember from the security point of view.

A

Yeah so yeah, okay, yep thanks yeah, so right now, I think there are a couple of uh like caps we're trying to move forward. They were actually you know, actually see a lot of things that we want to move forward. Right. Cpt is a big one and there's also the boarding group one that are there any more questions about CBT cup, if not, maybe I just moved so that I'll just say there are like a few of those.

A

We are actually uh trying to move forward in 1.26 release.

A

um One group right yeah, that's why this is my my cat, so this one has been there for for quite some time, so just want to make some progress, but I think we yeah I, think we need uh with this. One still need reviews actually still need the API review and also I have a I have a PR in the container search interface, the the CSI spec repo. That's also need more review right, so this is still this one's still pending. Hopefully we can make some progress um yeah.

A

That's this one and then the next one is um Warrior mode conversion. This is the you know. The security issue with regard to, if you change the water mode from file system mode in Roblox mode, then you can retrieve the change blocks. Then there's some security concerns right. So that's why we have these features right now. This feature is already Alpha, um so we're trying to move this to Beta and then we, of course we hope uh you know uh backup vendors actually leverage. This feature.

A

C

C

I mean to support I mean once we have CBT to leverage that CBT to backup recover file system. Please you will need this conversion right.

A

Right even like do you do you guys do this today, even without the CPT like uh when you create a PVC from a modern snapshot, do you also try to change the mode? No.

C

A

Because no, we don't change it. Oh we don't we don't. Okay, you just basically use the original file original mode. Okay, but.

C

Yeah I mean like unless, like Ford F series is different, but yeah like once we have CBT, then it makes sense right. We can change the.

A

Model, oh I, see so you're thinking this might be dependent on speaking, okay, yeah.

F

So this I have plans to change mode for restoring a CPT in an infrastructure which doesn't support CBT directly. You know you pull this. You pull block data from uh off, you know from the repo and restore it in a cluster which doesn't support block mode yeah. How do you do it? You'll have to you know, do it as raw block, but maybe the PV, the application you're restoring, was actually using file system mode in the PVC.

F

So you have to then you know, transform that pvpvc pair into a another pvpvc pair with Master mode, so that conversion is critical, but.

A

You do you do that already today. Do.

F

The conversion already uh no I, don't have the support. Yet this is in the in the roadmap. Okay,.

A

F

uh But you know it's that it is pretty right.

A

Right, so this is basically like a security thing right just to make sure that it's not exploited by malicious users. This feature yeah this particular thing: it's not. It actually tells you. You know in the volume snapshot content that tells you what is the original mode, so you can actually compare yeah.

F

Okay, as long as it doesn't stop us from doing it in backup software, it's all right. No.

A

No, it does not stop you, it's just making it more secure because of the back cover compatibility. We actually still. We cannot really force you to you.

C

Know yeah I mean directly. This gives blessing right. That is allowed.

A

Right so basically you know, if you you don't use this one I mean you can still go the way it is before. But then we just you know, tell you that you could have some problems right. So it's a you know. If you, you really need to convert it, and you should actually use this feature just to make sure you know it is allowed. So.

B

Next, sorry, um is that part of the a core kubernetes API the volume mode conversion, or is that part of CSI.

A

This is your CSI right. This.

B

Is still yeah, this is not the PVP. It's not going.

A

To be in the external provisioner and the snapshot.

B

Oh okay got it okay, so he's not at the PVC, um not.

A

In court, no no.

C

A

uh Yeah, so this one basically we're trying to move this one to to Beta, so there are just right now there are some I think one question remaining for the for prr review. Basically, it's just a question about the Matrix uh other than that um I think. The other thing is the E3 test, so actually not sure if we can get that in before tomorrow.

A

So there must be some questions on there because for this test for the EQ test, we're going to add ET test Auto tree uh so Yvonne, that's another comment that I added on your cap, because the CVT feature is completely out of tree, so E3 tests I would suggest that I want to be added out of tree same as this one same as this one. So once we have this framework, then we can do similar things.

A

So this one is kind of trying to establish a framework because right now we don't have anything any test that is in the kubernetes CSI people.

B

Oh so right now we don't have any e2e tests.

A

Well, there are tests, but there are not tests that that we say there are not no test code, that is in the external repo, all test code or entry. But now now what we're trying to do is to um for.

D

The new test well.

A

I need new feature, you know if it's like Auto tree feature I think make more sense just to add those tests, auditory yeah. So.

B

Currently, all like the tests for say, volume snapshot, they're all entry is that.

A

Yeah, all women's natural tasks are entry. Okay,.

B

So we're going to move the rf3 okay.

A

I think that makes makes sense, but still you still have to be added to the you.

E

Know so your body.

A

Is testing for yeah? That's still the common thing right, but but the test itself will just uh add an auditory, but basically you need to have a framework right. So that's that's. Why that's why I actually take some time for us to set that up so uh ronak who's working on this cap, so he is a working on that. So, let's see hopefully that can that you know we should be able to get that one in some time. Even even if we couldn't make tomorrow's deadline, we should get that I mean sometime.

C

Later, okay, okay,.

A

um So that's this one.

B

So yeah yeah so sorry platform. The cap perspective there's nothing the from the cap perspective. There's nothing specific needs to be added right, like because I already edit, like some of the basic e to e test case. So I guess, your comment is just a. This is just molecule FYI that it needs to be added out of tree yeah.

A

I think it's okay, I think it's okay cool! Maybe you have to.

B

C

A

B

Okay, cool sounds good.

A

And also because yours is the alpha, I think it's like for beta. We actually need to have the test in place merged showing up in the test grid, since yours is still Alpha.

B

A

um It's basically just need to have some. You know have some description saying you know how the test, what kind of tests you add right.

C

A

Yeah yeah, um okay, so I yeah I, think those are the main three main things there.

A

Anything else you guys want to discuss today, oh so, for those who joined late, we actually talked about the two cons and there's a uh there's, a contributor Summit on Monday.

A

It was also a contributed, meet and greet on Friday and also a data production. Queen group has a session on Wednesday. Six sword has a session Friday, so I just have those information here. Thank you guys are interested where.

C

Is the meet and greet.

A

uh I can put the yeah I can put the information here. I will find that yeah I think that yeah this this should be like for anyone. If you are in there, you should be able to go there, but for contributors I mean, do you actually have to register and also you need to be like you know, kubernetes or member? If you are not, you still want to be there. uh I can sponsor you just when you you know when you fill out that form. You should put me as your sponsor.

A

That's the contributor Summit, but for me the contributors meet and greet. This is when this is for anyone. So this one should be fine. Anyone should be able to go there and then also the sessions where anyone can go to your sessions.

E

I have a question, is the is the Sig storage session different from what I heard was the Sig storage meet and greet? Is there yes, I.

A

Mean the difference yeah that uh meet and greet okay, so this is a it basically just a more for the new people. If they're you know they want to know what is extraordinary, they want to meet some people there right, so um so there will be like I will be there. I think there are. There should be a couple more other people who are who are there, but then there are also new new people who just walk around want to see. You know how they want to get started.

A

They may want to go there and have a chat. So this is like a casual thing.

D

A

This is just casual Gathering uh and the sixth story session. This is a formal session. Let me show you.

A

Right, so this is so, if you go here, if you click that link, it will show you right. So this is six Soldier sessions, so this is the kubernetes six or deep dive session, uh so I will be doing the session with uh uh Mauricio. So this is basically talk about.

A

We basically just give an update for what six storage is doing in 1.25 I'm going 26 release. Mauricio is going to give a a deep dive on CSR windows.

A

Yes, this is like a presentation. Yeah.

E

Yeah, my my colleagues had alluded to a Sig storage, meet and greet as well, and they were wondering if there was going to be one this year at kubecon, Detroit.

A

Oh meet a group okay, so um so meat agreed. Is this one? Okay? Would they wear at La? Okay, when did they go? It was the last.

E

Previous years, so yeah I I, don't.

A

Know exactly the Valencia okay, I guess my question is uh I I wondered so this contributor meet and greet this is for every sake. That's.

E

Everything overall yeah.

A

Okay, yeah yeah, so every city, if they have any uh representative, they want to be there to meet people they can. They can sign up right, so I have already signed up so so I will.

E

Be there okay, so yeah, so that effectively works for it. Okay,.

A

Yeah, let me yeah I will put a location here as well, so there's a location well, actually, I, don't know if they have added location, but but anyway, when that's available, I will put that there and also this should also be in the like the uh agenda. I think when, when you analyzed it should be there yeah I'm.

E

With you, so so it suffices. Okay,.

A

And also I think last time in Valencia, which is the in the kubecon Europe uh last time right so I couldn't go because I got a Kobe, I canceled it, but we yeah, but we actually had a. uh We had one. um We actually we actually had a uh sick storage session. Just in here like we have like one hour thing. We basically requested the table. We have one hour there. Yes, I did not go there, so young was there. So he was there.

A

I didn't request you this time, because just uh uh I there's still not that many people attending so I did.

E

Not okay and you have yours, Instinct, storage, session right and.

A

Yeah we also have yeah. We have this, one I think it's sufficient so, and this is more like a um so they will have some some uh I think people can submit a um session proposals for this one as well. I think they're I did not submit because, um because I actually have two I still and there is a data on kubernetes day, I I have a panel. There I need to go there as well. So I didn't submit this one, but I will be here in the morning. Be there in the afternoon.

A

I probably will go to this other place, um but but there are some tables there right. So people. If people are there, they want to meet there. We can still chat. So it's just not like. We don't have like a sign saying this story, but it's more like a unconference right. We can still do that yeah. If anyone wants to be here on the chat, we can still do that, but this one definitely you know we will have a we'll, have a sign, saying, there's six stories.

A

If you want to see other sick, they will be there as well. That's this one thank.

E

A

So you will be there yeah.

E

Yeah yeah cool.

A

Okay, uh yeah, then yeah so come to this. One also I think this actually happens before the secretary session, so this one's the 12 30 and Eastern time and then six storage session starts at two okay at 2 p.m: eastern time, yeah and data protection. This is on Wednesday. The session is on Wednesday.

E

Perfect I'll follow up on all this. That answers all my questions. Thank you.

A

B

So Shing I send you the recording for the CBT update for the.

A

Oh great yeah I will include that in the sixth recession yeah. Oh.

B

No, actually, nothing sorry working group yeah just take a look to make sure the file is working on your.

A

Computer, okay, I need yeah sure.

B

I've recorded it on a Linux laptop I just want to make sure. Like you know, it's.

A

All right, I hope that should be same right.

B

Yeah, it should be, it should be all right, so just purchasing.

A

I will check I said. Let me.

B

A

Play that and let you know: okay, cool, okay,.

A

And okay, uh anything else.

A

Right that you have a few minutes back thanks, everyone thanks.

E

Everybody thank you. Bye.

C