Kubernetes Ingress Nginx Subproject, 16 Feb 2023

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: [SIG-Network] Ingress NGINX meeting for 20230216

Description

[SIG-Network] Ingress NGINX meeting for 20230216

A

Hello, everyone today is February 16 2023, and this is our cncf Sid networking subproject meeting for Ingress nginx, which means it is a cncf project and it adhes to the cncf code of conduct. If you have any issues or if there are any problems, please report those to myself or Ricardo and if need be, report, those to the same network leads, and with that do we have any new folks.

A

It looks like it's just the four of us hanging out so I, don't think, there's anyone on the phone, so we can just roll right into uh review on some of the at times, I restructured the agenda a little bit, I didn't like how it was so I just set it up so that the agenda's here so I want to make sure I, don't think we were. It was on the docket make sure we review previous week's uh action items. Mostly.

A

It's the keep myself in check because a lot of the action items I assigned myself so making sure that we're reviewing the action items and then the issue triage I, also added the board back in there, because the issue triage link goes to the issues board. It goes to the issues. This actually goes to the project, board, open topics and then any new action items making sure that we've got those collected and added um so yeah.

A

So with that, um we had some issues with our last couple of releases: I, don't think, there's any real action items to do on there, except make sure that um our end to end does what we think it does. um But uh we do need to pick up the path type, so I'm going to add the path type stuff. There um just have a conversation with Ricardo and myself and figure out like what was the actual issue and make sure it's not non-breaking. So I'm going to put that on there just to schedule some time.

A

A

um So we'll schedule some time we'll drop in the channel. If folks want to um hang out and discuss it with us and just review the code that was there. So we can try to make sure it's a non-breaking change and get that back in there um were there any other. So I have the otel one here um and I I'm, not a hundred and Sean I'm glad.

B

You're here I'm, not 100, I'm.

A

Where we're at with that, um if it is actually all fully implemented, um so maybe we can discuss that when we get to the open topics, um because I obviously did not create a ticket to remove the other room of Zipkin and Jaeger. So that was that and I just want to go back one more because I'm sure there's other stuff in here: egress um regex for that stuff. So that one was the issue.

A

um I didn't do the write-up for the alpha release of cpdp, um so I'm gonna put that back up on the action items. This is just like I said this is a lot of this is mostly just to keep myself in check. um So I'm gonna go ahead and add that back there.

A

Anything else that may be missed um on the action items that we know that we need to review or um catch up on. I know, we've been it's been a month and it was the New Year prior. So um anything folks want to discuss before we uh start during some issue. Triage.

C

Nope I uh I, wanna Focus next on finishing the cpdp stuff, but I don't think we will be able, before releasing at least two breaking changes right so uh yeah so I've been working today on the uh reviewing some low hanging, fruit, PRS and merging those as well I want to kind of try.

C

At least you speed up some some PRS that are on the QE for a long time and if we can say for sure that they don't bring any packages, merge them, because we have a lot of those that are just waiting, someone to say: hey, okay, this is fine. Let's, let's merge it.

A

Yeah, there's there's a lot of a lot of just low hanging fruit ones in there um and I've been trying to get those in um as well like little things.

A

um There's one in there like dropping Ingress class from the the cluster, not from the cluster role from the from the role which yeah it's not needed. It's in the cluster rule. So little things like that agreed. We need to get the small ones we need to get through faster, um just because yeah and hopefully be able to tackle the bigger one. So I agree all right um cool.

A

Well, let's jump into the fun. Do the issue triage. So, as I talked about last time, we discussed it. um I know this is probably kind of small for other folks, but I'll click through them, as we do um as new issues get added. They get added to the project board with no status um and then I don't know who has privileges to add things to the board.

A

But if you go look at an issue you can add it to add it to the board and if not ping me and I can add it to the board. So I'll show you what I mean I, don't know if you've all used GitHub project.

A

So if this is redundant for everyone, just let me know I'll just do it for the recording, but if we go back to like older ones, because I haven't added new issues to the board, so we go look at one of these, that's assigned to me and just go ahead and add it to the projects there and then it will it'll give us a status so right now it's no status.

A

I'm gonna drop it into do because I've already assigned myself to it and I haven't looked at a lot, a lot of the AWS there's like four or five AWS issues out there with like albs and LBS and I, don't know if they're grouped together or if they need to be but um anyway. So if you need to do that, click the button, add it to a project and it'll drop, it no status, and then we can review it during the triaging. So.

D

A

Yeah refresh the board, it should come up now into in the to do um yeah right there. It is so try to track what we're working on who's working on What's. um So folks ask why we aren't. Why are you on looking at my PR or anything like that like here? It is I, think you can also add PR's to it.

A

um That'd be a good thing to know if we can actually add PRS to the project board.

A

So let's look at this one I, don't think it automatically adds PRS as well, so yeah um probably have need to have a GitHub action that automatically adds the PRS to it as well, so make sure we're reviewing those as they come in in a timely fashion, all right, reviewing them or assigning them out to folks just again trying to get better about moving things through the system and getting updates in, um but that's in progress just waiting on him to update those.

A

So anyway, that's the new process on how we're trying to review things and get things moving through the system. If you have issues adding things to the board or you don't see something, let me know um like I said still working out on the permissions on all of that.

A

uh So with that, let's just go start it I, don't know if it goes from the top to the bottom. Let's go. Look so 9403, don't know what order these go in 96, so yeah, the newest or at the bottom.

A

So you think start with the no status and go from there.

C

You are asking me, no yeah. What do you think it's better I trust.

D

A

You guys read that, or is that too small.

C

uh Can you increase just a bit yeah.

E

But I can follow along so go to the actual issue.

A

All right, it says the support question. So, let's see, if I can actually do it, I've not read through it. um So it's 94 or three for the folks where's the chat box, I hate. How Zoom does.

D

A

That's the one we're looking at.

D

A

Chart I thought the internal one I thought someone added the internal load, balancer IP.

C

I uh yeah, so to be honest, I I'm, not aware of anything of how those Cloud stuff they work. uh The deployment on AKs gke whatever so I cannot give my opinion on those ones.

A

Yeah, but we can still give them the option in the helm chart to add in a load balancer IP to an internal, because we the options in the helm chart because you can do internal or external or internal and public I. Forget what.

D

It's called I'm.

A

Guessing it's just missing the the load, balancer IP. Let's go look.

D

Let's do a quick look real.

A

uh It's values.

A

Not service internal, ah it doesn't have load, balancer, IP.

A

And that is a real low hanging. Fruit, PR.

A

D

A

D

The custom template.

A

Oh, it's coming it out so.

A

Did he not change the value.

A

A

C

A

Did see this, so this is the same issue that I did sell from the pr perspective.

A

Thing is that it's already there I'm confused.

A

On the surface, it's already there.

A

's a good question: he asks um what is the what's the effect of commenting out something in the values at yaml. Did anybody know off top of their head? I? Really don't because it is commented out here, but it's not in the dock, so I agree. It should be in the docks.

A

I can take it on as an action item to answer his question here. If nobody knows off the top of their head.

A

Cool okay, I'll. Take it as an action item, follow up.

A

Yeah go to the next one.

C

Go back to the board.

A

And I'm assigned, as that one am I assigned to the issue as well I'm assigned to the issue as well yeah. So it's on me, I just haven't looked at it. So that's good good to know.

A

As I'm going through these, are there any issues on here that people want that? We should be talking about this. One doesn't sound good, I read through this one Ricardo. Did you get a chance to read through this one, no matching key in nginx and local store, no.

C

This is pretty old as well right. One two one.

A

December 22nd I.

D

Mean it's a release.

A

Oh, the release, yeah one two one is a little old.

A

We can always do that. Ask and see if someone has asked them to upgrade yeah. Maybe.

E

James, if you want to assign that that one, that nine nine four zero three to me, that that'd be fine I was sorry. I was just catching up with the conversation. I had to walk away for a second um I I'm, pretty well versed in Helm and Helm charts. So I can answer his question. Okay,.

A

E

In late on that, one, but not.

A

Talking that's.

E

One I can help with.

A

Yeah I'm trying to figure out why why we have one commented out and not the other, because.

D

A

Internal is also uh we don't always deploy internal. We always deploy this one, um but we always don't employ internal unless they ask for it. So I'll go ahead and I'll sign this one out to you.

E

Yeah I mean I I have my that's one. One area I I do know quite a bit about because I wrote the helm chart for ranchers deployment of our stuff. So I can take a look at this and see see what what the deal is. Gotcha. What's your name, it should be just Dylan, Dylan Turnbull, all one word, uh I, don't know if capitals matter, but it's capital, D capital, T,.

A

I wonder if it's because you're not in the org.

E

C

Yeah you, you should probably go to this uh issue even and do a slash assign for yourself. Okay,.

A

Well, let's not assign also assign it to him because he's not in the organ.

C

Yeah, but at least we are gonna, know that he assigned to himself.

A

Yeah, okay, that's fair um Dylan! That might be something too you might want to go. Do is actually go put in the sponsorship request for the kubernetes org; okay.

A

Okay, sorry back to the back of the next one.

B

um Can we talk about 9606.

A

C

A

Car you wanna, you wanna, take a look at this one or just ask him to upgrade and see if it's still an issue.

C

uh I can take a look I'm, not sure. I'm gonna have some time for the next few weeks. Yeah.

A

No, that's fair! That's fine! With the dpcp um stuff I wanna I want you.

C

A

C

And then moving houses so yeah.

A

Moving okay, I'll I'll follow up on it because that's an older version too.

A

Okay, um sorry long I wasn't ignoring you. um What was the issue.

B

uh 9606 Ricardo somebody is creating a headless service and then they're, creating a manually creating the endpoint slice and the.

D

B

Address in the endpoint slice is not even in the same name space it's either in a different name space and one person is actually creating uh using IP address outside the cluster. So is that a valid? uh Should we just tell them? That is not something that we support.

C

Well, this is, this is valid for endpoints as well right.

C

Yeah, if you create a service without selectors, then you can create. Then you can create an end point uh with the same name of the service. This endpoint is not going to be managed. Then you can ciao Ingress to point that theme. So uh uh another question yeah it is. It is actually a bad thing to support, because someone.

B

Can just use Ingress.

C

To by Pastor, wait.

B

Wait the question is uh second question is about doing it that way, but the first question is: the destination is not in the same namespace.

C

Yeah, so so that's that that's that's the point long. So uh we shouldn't, but even today, with with service I guess we do support right, because when you, when you create so we manage our back our back ends based on the end points are not on the service. You can create a service without a selector and then create an endpoint in that same name and tell Ingress, hey, use this thing and it's going to use right so yeah, it's actually bad, because you can use Ingress to bypass some rules.

B

Yeah, that is why I'm asking and more specifically, the very clear requirement two two people are chiming in on this. Is that the part to which? Finally, the service points is not in the same namespace it's either in a different namespace from where the Ingress object is, and one person actually creating a destination IP address outside the cluster completely outside the cluster. Now then, yeah.

C

This is this is this is why eight points and services without selectors they maybe use it for the problem- is that today we don't manage that, so we we should do away saying that if we select a service that doesn't have selectors, we are not gonna even move forward to select the eight points of that theme.

C

B

Because, okay yeah.

C

But uh we need to raise an issue on that uh uh say that we need to implement this protection I'm, not sure if this protection is in place. This, in fact, can be a cve uh because you can use Ingress to bypass uh like Network policies and other themes right.

B

Right right right so.

C

You just need to know you just need to know the part IP of uh of the other side right as and as soon as that power IP doesn't change. You can consume that, so this.

D

Is a problem yeah.

C

And uh we should take care of that for sure. So, let's open an issue for that and we need to work on the reconciler to add a flag say that if we don't have selectors, we are not even going to move forward to the eight points or the eight point slides selected. Okay,.

B

Can we just continue on these 9606 because it has got clear uh requirements from the user explicitly mentioned there.

C

Requirements like.

B

uh He created endpoint, slides manually and it's a headless service and the IP address the destination part is not in namespace, and one more user is joining in there and saying yeah. I also did the same thing and he's getting there about getting a crash and panic, and then the other second is actually Point wanting to reach a destination that is not even in the same cluster. So if we update this.

C

Yeah yeah yeah, so so long. This is this is the case and we need to so previously the way that people did. That was creating a service with the external me remember, and then uh we we supported external name, and then someone figured out that external name could be used to bypass some things like going to localhost or going to somewhere else and bypass.

C

This is this is something really similar to that and in fact, any any Ingress, uh any Ingress that uses the same approach of the end points and not checking if the services contains a selector can be, can have this problem right, so uh I, I, I I, don't think that I'm gonna be able to read and pass through all of this right now.

C

But uh this is important, uh so I would I would kind of add a priority version for that one James and we need to, and we need to take care of this one, because this is a security problem actually- and this may be a security problem for for other Ingress implementations as well. Right.

B

Exactly exactly so, we haven't made a statement that it is not the right thing to do so.

C

Because it's not, it is supported. The thing is that we never added also a flag, saying hey, we are, we are not gonna, allow this or that case for the administrator right. So yeah.

A

That's causing this bug right here is that we're not checking if the endpoints are there, because.

D

A

The same, it's the same spot in the code, so there this person's pointing out that it is a bug and we panic when it's not there. This person is trying to use that functionality, so we both need to one not allow it and we shouldn't Panic. We should just return an error and saying there's no endpoint slices on this one because we're looking at it from a selector's perspective. So both of these people are right, they're, just using it for different means.

A

So I don't know if we should accept this one that just run that just runs through anyway yeah.

E

A

Don't know if this this would. This would enable that six would enable them to do this, but we don't want to enable to do it. We want to actually stop it because it is a security concern. My understanding is correctly yeah.

C

That's correct.

A

Okay, so I will mention it in 955 that we actually need to I.

C

Mean hold on James, so this this one. So this one it's like this one is it's kind of okay, specifically this one, because we are checking a pointer this this PR, we are checking a pointer that may not exist right. So that's that's that those teams that we should always be like hey is this not new, and if this is not new, is this ready? So this this PR? It's fine but I!

C

Think in that long it's pointing and and the thing that that we need to to figure out right now and I I wouldn't merge that right now. It's because we have a security issue that that people may be using a a service without a selector and then creating endpoints to bypass right.

C

So as soon as you know, the IP address and we kind of block at that for localhost, because we do all of the management of of the back ends and so on, Via localhost, but you still can use that to bypass uh and go to uh other pod, IPS or even kubernetes API right yeah. So this is. This is kind this.

A

Is problematic.

C

A

So yeah, okay, but I the pointer issue. We can fix this one. We need to say no. This is this opens up to a security vulnerability. We should not we're not going to allow this functionality and we'll open up another issue to fix this hole, but we're closing this is not a bug.

C

uh Yeah, the paint coin is a bug this one because we are not checking the pointer right, but we have another one which is actually a problem, a security problem. So we need to open an issue for that. One.

A

Yeah so I'll I'll go ahead and do that.

A

Anytime, that's fair! That's good! Okay and.

D

Then we can accept this.

A

A

Go ahead was that you long.

B

Ch rooted image, somebody wants.

A

B

A

Random yeah I saw that one. Sorry, there is what Sage root can't reach Dev random.

C

uh That wanted some a long time on my QE right yeah. Do we have a fix for that long? Did you solve something that could be used.

B

No I, just uh it's a it's very simple problem description, I, think easy to understand, but I don't know which way you want to go, because if we, if we give Deb you random, then there's no point in having sex with image but for some reason he wants to access preview. Random, slash Dev should not be accessible at all. uh If you're doing, if you are serious rooting at all right.

C

B

On this, just which way you want to go, if you mentioned now later, when you are free, then we can address this yeah.

C

I need to think on that, because I I, honestly uh I'm not sure how much we are going to keep the CH root or if we are gonna just as soon as you do. The split control plane data plane if this is not going to be required anymore right. So it's like if this is something that is just for one user and it doesn't generate a huge impact. I would just keep this way. Otherwise uh we need to fix it, but so to be really honest.

C

I I, don't know when I'm gonna have time to look up at all of those teams.

A

Got it okay, okay,.

D

Well, we looks like we've.

A

Talked about this before yeah.

B

But the the split uh cpdp split was not in too much in context, then, but now we are closer to that. Cptv split implementation, so if it could go away the whole con, the core concept itself of CH root is going to go away. Then this is mood. I mean this. There's no need to discuss about this at all. Yeah.

A

No one's again, neither like I'm trying to understand what the use case is why they need to reach Dev random.

B

He tried to run something of something that he wanted to do he's.

A

Mentioned it in a ticket.

B

And I'll check.

A

B

Create a file or whatever amount something about.

B

D

Mean the point, is we.

B

Are offering CO2 image and it's valid- that somebody wants to access something logical anyway,.

D

Not not I mean we know now.

B

That wait for CPU to be split. You know which way is going to go.

A

This is one of those ones where it's again, if, if they need to support chorus, we we can't support all of them across I mean we've got all the ones that we're testing and supporting. Now um this is a question of a community need. Is there more people? Are there more people using chorus fedora and if there are then yeah sure we need to fix this, but right now it's only two folks and we need to prioritize the other things that are in our two um and that's unfortunate.

A

But that's how that's where we're at right now, yeah but I mean it looks like um Ishmael is looking at it and if he can get it working, that's great. um So he's already got someone assigned to it another question of: if we should do it, that's a different one. Technically, if someone can solve the problem, that's great, but should we allow them access to? It is a different one. So I think we wait and let um let Ishmael figure this one out, and then we can have that discussion um but yeah.

A

Hopefully, the CH root is going away as soon as we get dpcp, and so that's why I try to guard Ricardo's time on issues, so he can get that finished anyway. Okay,.

B

Ricardo uh Essence.

B

Are ready for merge, I think they're, just waiting for approval James, if you and Ricardo can just look at it, then well.

C

As soon as you as soon as your folks tell me it's fine to merge, we can merge I think that I would just do a final person that maybe I would like a sound to to make like a demo for everyone on this working on this Branch just to say, hey. This is how it's going to work and even Point people to how this is gonna work right and then we can just get rid of that. This is my my take on that.

B

Yeah I think it's ready.

E

B

E

In yeah it's ready and there's a repo, for example, so people can try it out.

B

Can you just mention the pr numbers.

C

I've seen that today, I guess it was open on September 22 yeah this one. So this one this one specifically is on my list, so uh I I will I will take a look into that. I! Think it's it's fine and it's it's actually great to have this measured. Finally, uh I just want to take a final look and see. Okay. This is fine.

C

Let's merge this and then, let's uh start pointing people to use open Telemetry, instead of any of the things on Zipkin, Jagger whatever and then, let's start removing Zipkin Jager from from the call sounds good.

B

Yeah sounds good he's, actually, Essence is actually written in a small documentation. Also I think very small documentation example. So.

D

I want to write that.

B

If it is merged, yeah.

E

There's also a PR for documentation.

B

Can you mention that or that.

D

B

D

A

B

Maybe yes, maybe you can filter by Essence name yeah.

E

It's nine one, four four.

A

A

So Ricardo between now and the next, our next meeting he's gonna review and look at it and then Sean do you want to do be prepared to do a demo on our next community meeting and then we can get emerged.

A

Okay sounds good and by that time, we'll probably have three releases done just kidding all right.

C

Broken weren't working.

D

A

um Okay, so we've got those both on the projects. um Do we have any other issues we wanted to discuss because I know we've got we've got time we can keep going through the triage ones. um I'm gonna. Add the open I'm going to add that to an action item of um Ricardo you're, going to review the hotel PRS and it's on you're going to do a demo.

D

B

uh Another quick question regarded: did you change uh health check probes from TCP to http and https anytime in the past few months,.

C

um No no I mean I, don't remember what was my breakfast long, but as far as I remember no.

A

I I know which I'll pull up the issue. There's a health check, yeah.

B

Go ahead, James, sorry.

A

No, it's fine. This is the this. Is the default one right yeah, so.

B

People are complaining. Multiple people are complaining that on AKs on Azure only.

A

uh Yeah there are so I've noticed that too there's been a pickup of uh AKs issues as well as uh AWS.

B

Yeah- and this is the significant one out of box install, they are saying, um doesn't work at least three people have said: I think it doesn't work, because the probe changed from tcpe to http, https I, don't know if it's recent or not, but it's not.

A

B

Very clear: okay.

A

Okay, this is this new one.

A

Oh yeah, no, it's this one 9601.

B

Yeah, that's that's one and then.

A

I don't know about this config map server, snippet.

B

That is, uh that is odd, because he's he's doing some funky uh directive conditional directive that there's a if, if there is a header coming um in the.

E

B

And and if it is coming from the Azure uh lb type of service, then it it kind of bundles, a header automatically and he's checking for that header and he.

A

B

A

B

You and down below he has been provided some links as to where some blog, you know did that, but.

D

A

B

A

Is a self-induced problem then, because we can't hey, we we provide the health checks and he's stopping the health checks from being from working because he's only accepting traffic. So he needs to add something into the server Snippets for, like the location on slash Health, to accept any traffic from those load, balancers or figure out how to get the AKs load balancers to send the right headers.

A

So I I, don't think this is an issue on us uh at least at surface value. Just when we're reading this and understanding as anybody else object, I, don't think this is a bug. It.

B

Is a configuration so I told him to do that. I told him to find a way to allow health check probes as well, but I mean I am taking the time here. Is this guy and I? Think two other people said that the probes changed from PCP to uh https, so I asked if it is. If we are not sure that we made a change like that, then it is not a big deal to talk about here.

A

Yeah, no, that's fine! If we change the probes and we introduce the problem, then yeah, but we can always just that's the nice thing about git. We can go, look and see um also, it didn't say which way they're deploying this.

A

Let me go look. Did it say how Okay so he's doing it? Maybe at home, so.

B

I'm not copy. Well, it's obvious. You see that the health check probe goes on a board and the end point, and- and it's just DPS for sure, so he uh I just wanted to be sure that we didn't change from TCP to https. So that's.

A

Can I not didn't get women there.

A

Was it not on the service where's where's? The health check at then should be on the controller service right.

B

I think it could be more than.

A

I'll be on the deployment. What am I talking about?

A

What's kubernetes.

A

B

A

All the MLS will only have.

B

The uh listening part of it, not the probe part, the pro part I, think- will be in a go file right.

A

D

Going to be on the deployment.

A

S Readiness problems here they are this, so they're defined in the controller they're defining the values live lens proposed. Sorry, folks, I'm just saying: if we can close.

A

See when I broke it so well, I only made those. uh This is gonna, be because.

D

A

Did the refactor with the uh the values and the helm Library? No, we're not gonna be able to know if I have to go back a little bit more in history. So we're not going to be able to know right now. I thought I would be able to see how this top ahead, but we're not defining it. I, don't know, what's the what's the default, if it's TCP, if we don't actually because we're not, we don't set it here.

D

B

Any case values file will only contain a spec for The Listener. The health health probe, listening endpoint um I think he's claiming that the actual probe itself.

A

Yeah I'm not saying that.

A

D

C

A

Should be good, oh again, because it's uh we'll have to look at it. um You know because it's HTV here, it's HTTP, we'll have to see what it was before. I made all of those changes to Helen, but again I. Don't I, don't remember changing it, but you know there's a lot to come through.

B

Okay, no, if if it is not, if it's not a clear collection, if you can't recall clearly let me change the protocol, then it's no.

C

I, don't think it is.

B

Our problem, I, don't think we changed that or anything of them I think he's self-inducing. The problem I just want to confirm that that's all.

A

Yeah, even before my change, I don't see any changes to the the probes on things have changed. The probes and yeah I think it's a misconfiguration, not a misconfiguration. It's just configurations that just don't work together: okay, okay,.

B

The next topic- I just wanted to get one one liner from both of you, is I, think we are piling up uh issues on people asking questions on cars. I, don't know how do you pronounce that c-o-r-s cross the region request course: yeah cool, so um I Tried reading and trying I tried to understand enough about it, but I think much deeper knowledge of HTTP and course is required to earn things. I think there are several issues uh for a long time for I.

B

Think six months we have in updated or provided any feedback to people yeah.

A

13 open issues that mention course.

B

I think last year one country meter from the U.S was active on one of these issues, but then he went to a he's not been around for a while. So yeah.

A

This one's, not even us, is it three three Thirty Four, oh, no, that's us and what is this.

D

B

Changed something in you know a course of the way we handle course I. Think we hardened it or the two people got in game stuff in the court for manage managing course um and now, but basically there's many questions unanswered.

B

About course, you need.

A

Yeah unfortunately, I'm in the same boat, I'm, not I, I, wouldn't even I, know how to spell cores and maybe pronounce it um Dylan. Do you think there's anyone from nginx who can maybe help us look into some of the the course issues that we have.

B

Well, that could be fantastic. That can be fantastic, even if they give just some kind of tips. Also, it will help.

A

He might have stepped away.

D

A

Reach out I'll ping him in slack and we can take it offline.

B

Okay, the the way I'm bringing this up is I. Think, six months ago or eight months ago, uh we made a change to or we we basically merged, somebody's PR it does not. Ricardo is not until now somebody else a contributed from the U.S came in, and um we.

D

Merge their PR.

B

Ank and so now, basically people say that if they, if they use start um as the origin and then they do set requests for auth, so basically people are having problems with uh art. You know art URLs, reporting back coming back and oh.

D

A

Yeah, um okay, if you can find, if you can find that PR that you're talking about where we did the hardening on cores that probably helped too um and then I'll ping Dylan offline um and we'll try to see if we get someone that could help us out with that. Because again, I don't want to pull Ricardo off of cpdp I! Don't have any background knowledge on cores and um yeah.

D

E

A

See we'll ask if we can get some help and go from there? um Okay! Well long! As always, you've got you keeping the pulse on all of our open issues. So I appreciate that. So thank you for that um anything else. We need to discuss folks before we break.

B

C

We just to make sure that we are all on the same page. We are gonna before I'll cycle back to uh cpdp. uh We had that conversation and I'm gonna do a another breaking change for the peptide. Okay, so uh I think that we need to fix those things and uh I just want to have everyone aligned on that.

D

A

I put a thing in here for us to do a code review and to discuss it to just have a separate meeting to discuss that. And so when we, when we find time we'll let everybody else know if they want to join.

C

Yeah, just just let you know that before I jump into the cpdp I'm gonna jump into that one at that other breaking change. Okay,.

A

B

Yeah, okay is this about the VM that we have.

A

Yeah yeah this is yeah. This is about putting the path type back in and um trying to trying to get it in, um but also yeah yeah, that's the past type, and then we've got annotations bits, so we've got a couple um fixes that we need to get put in.

A

um That's why these Community issues the the issues on GitHub are piling up. So we've got to get the low hanging fruits out. We've got to get the fixes in um for the path type and annotations, and then we can focus on the next steps. So yeah.

B

One quick remote give a quick note, I think when we try to fix this 163164 release things um the deployment page on.

A

B

A

The index.md didn't get updated for some reason: I.

B

Saw that issue this morning, yeah, so somebody sent in a PR to change just to uh I, requested him yeah yeah.

A

I saw I, I saw him, ask for it and if he doesn't do it, probably by like by the end of Friday, I'll, probably just push it through. um Okay, I.

C

Mean I'll make.

D

A

And push it through and close his PR, because that one it's a simple documentation fix that needs to get updated, um got it. Okay,.

A

We don't have, we don't have enough time to have this discussion, but we need to as part of the release, one of the we have a couple issues, and one of them is the uh the image being released before we do the documentation of said release and the quick fix that I came up with having the conversation with Ricardo when we were doing 164 was that we don't do the pr to the kubernetes registry until we do the release on GitHub, because that way we've got the documentation and the notes in place, but you can't pull the image until we do that.

A

um The the problem is going to be the release. Notes won't have the shaw until we get that build.

A

So that's one of the things we just need to figure out the timing and discussion on that, but I don't want to release the image until we put the notes out there.

E

D

A

We've found the the get Ops folks and people are just pulling from the registry and not respecting the change log or the release notes.

A

So we've got to figure out the timing, because we've got the three release: three places where we release information. We release it in GitHub as an official release. We release a Helm chart and then we also do the pr to the kubernetes registry. So our data for our release lives in three separate places and timing is an issue.

A

So that's something we need to thank on from our release perspective um anyway. So yeah. We need to continue discussing how we do our releases yeah.

B

So I I'll be online. If you want to work on it or have some tasks.

D

A

Today, I'm action-packed full of meetings on Thursdays, so.

B

Anytime I mean because yeah I'm just saying that I think it's super important, because I experienced that the charter was out there and all that was out there before the release. So anytime, you think we should talk uh whatever I can do. I will do.

A

Yeah all right folks. Well, this has been a good meeting. We've got a lot of action, items and follow-up. Thank you. Everybody for joining and uh we'll go from there see.

B

You next time, sometime later on a schedule about the.

C

um Yeah yeah I'm I'm go ahead. James.

A

Yeah, we'll drop it in the dev channel, so I'll probably be it can be either tomorrow morning or Monday morning, um so that you can join okay, I.

C

Will be back, I will be back from vacations next week, so maybe we can just try. I mean I must do around, but using my time to deal with the movie and other teams, yeah.

A

Yeah, okay, um well, I, haven't looked at my calendar, but we can just shoot for Tuesday morning to do an ad hoc discussion on pastime.

D

A

D

Good, we'll drop the zoom.

A

We'll drop the zoom in the in the uh in the dev Channel.

B

In the DM in the DM uh chat and maybe China friendly time, so Jintao can join.

A

Yeah is that a good time for him in the morning, uh I guess the earlier in the morning. We do the better form, because he's 13 he's plus 13. yeah I'll I'll, try to find it so that all four of us can join. Okay thanks. Everyone bye have.

E