Layer5 Meshery Development, 23 Feb 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Meshery Development Meeting (Feb 23rd, 2022)

Description

Meshery Development Meeting - February 23rd 2022

Agenda -
- Autogenerate docs through cobra genDocs. #5108
- Enhance error messages. #5096
- Introducing --port-forward in system dashboard #5053
- Add support for `x-kubernetes-int-or-string` in jsonSchema
- Fix problems with invalid hook call
- Merge functional tests?

Join the community at https://layer5.io/community

Find Meshery on:
GitHub: https://github.com/meshery
Twitter: https://twitter.com/mesheryio
LinkedIn: https://www.linkedin.com/showcase/meshery/
Docker Hub: https://hub.docker.com/u/layer5/

A

Start with the meter awesome, uh so hi guys welcome to the machine development call. uh Let's get started with a few introductions first, so I see uh aniket ranjan. I guess this is your first measuring development call. Would you like to go ahead and introduce yourself.

B

Yeah, hello yeah, I am a nikhil. Currently I am working as a software developer, internet says cloud and yes, I recently joined last week only I joined uh mesri.

B

So I'm looking forward to do some good contribution to the um to this organization.

A

That's great uh great to have you here.

A

Anyone else are there any more newcomers who would like to go ahead and introduce themselves.

C

C

um My name is kessie, I'm from nigeria. uh I just joined missouri. Last week.

C

I work as a software developer for john hopkins um university chicago from nigeria, I'm privileged and glad to be here and I hope to learn as much as possible and also contribute to the open source community. Thank you.

A

Wonderful awesome to have you here so anyone else who would like to go. I don't see any more newcomers. I think okay pressure.

D

Yeah go ahead: yeah um hi everyone, I'm pressures from nigeria. um I just recently transitioned into web dev, so currently on front end. I'm really excited about layer, 5 and open source as a whole. So I'm glad to be here.

A

Hey precious welcome to the community, so I think we're done with all the new comments. So let's get started off with the agenda.

E

uh First off we have, would you like to speak to your topic.

F

Yep yeah, that's it! I will share my screen.

G

G

G

And just hide the panel for a while and the controls yeah. Okay,.

F

I hope it's visible, so uh my work is actually based on the cobra cli uh docs generation, 2d library itself. So it's all started out in from yesterday idea was given by lee actually so he gave the uh link and he gave an idea about how about generating the cli docs for mercury using the cobra cli library itself. So, as you can see here, this is the cube bella docks actually, and this is actually generated by the spf3 cobra library.

F

So if you can see here, they already implemented the functionalities to generate dogs in the form of any file format like man, markdown, rest or yaml. So I try. Yesterday, I tried implementing the same in the measuring machine ctl, so I added some of the functionalities like with prepending the template command that to comply with the existing ones and yeah. It turned out pretty well and I had to add a tree first like inserting the functional from the so that the command knows like okay.

F

This is the available commands and the sub commands present in it and then the what the function does is it will generate. This is just for sample path. Actually,.

F

So uh if I give this give the command instructions like generate docs uh in the form of markdown, then it will generate so what I did this uh after running the command. It generated several list of markdown files, so I've created a sample once so after running. This is the result like the same format, but instead of manually generating the documentation, the cobra library text had itself and it generated everything like the title. The synopsis examples options.

F

What are all the options inherited from parent commands and even the link to refer from the parent command if any, and it will also generate like the footer like auto generated by spf 13 if needed it, this can be disabled. So I also refer the documentation and also yeah. The yaml also turned out pretty well good, but some editing is required, so I will take at it from after the initial review of this, so yep.

H

That's my path and yep. I know it's just a shot.

I

Aditya, I just to confirm all the generated yeah I missed in part. What you'd said would just do so just to confirm real, quick, the of the generated um output, yeah.

F

H

F

H

F

Yes, yes, this is the generated markdown file, but.

F

I just didn't edit with the real path, since I didn't have much time, but this is the uh generic output, what it will generate for every command and sub command.

F

So if it's a command, then it will generate like this.

I

So uh husseina and navendu and piyush what what so do? What do you any thoughts on this potential approach, so you're all familiar with how mesh free ctl is currently being generated?

I

I'm sorry measure cto docs are currently created this. If, as a detail, looks into this, the this is under the the goal of potentially auto generating the docs the goal, it really isn't to change the formatting of the docs themselves, either both in k in terms of their urls of the reference guide or necessarily of the formatting of the pages themselves. I suspect there'll be a little bit of formatting difference, but that even that could be um changed if needed.

F

um Yep that can be changed using this uh with the given string, format and radio. Given input ceiling, the function.

I

So navindu isn't here I don't think, but piyush and hussein any thoughts.

J

I joined literally so the documentation is generated from the.

J

Commands a code right, yeah and.

J

So how I mean, which particular section you would be looking through in the code? Is it the init function or like the.

F

J

F

Yeah, I get your question so to generate the documentation like I had to refer some uh articles and then it's like uh there must be a main function which would call the uh gender generated dogs from the cobra cli library, and then only it would work so for for a while, I have created a sample file which is outside the um cli root dockets. uh As you can see here in the uh file three, so I've created a go file named gen docs.

F

So what it does is that uh there will be some functions like prepending, the uh markdown templates. um If I didn't give this template means the output won't be the same. It would be just like without this, so that there is an option to generate the customized template. So after giving the constant functions- and this one is for yamaha bits, but still in work in progress- and this will there are some functions like prepending, the header and the link handler.

F

So this is the main function like uh calling the three part, so the command needs like uh needs to know what are the commands present within the cobra cobra cli, so here in this case, uh I have to create a function in the uh root file of the measuring ctl, such that it returns the available commands and the sub commands like say it's a tree of commands, a root command below that it will be generating a tree right.

F

So it needs that so after that only it will generate the markdown tree and if the file has been done then, but for now I'll create deleted some of the files and kept only a little just for showcasing the output else. The ci test would fail to avoid that I've created a sample one yeah.

J

Okay, so, like uh let's say, uh system command is there and it has uh sub commands beneath that. So.

K

J

Those things you have to list in your code, which auto generates the documentation right. So that's all one needs to do right.

F

Yes, um or else like uh one needs to identify, which is the root command here in this case, is just the measuring ctl uh root. If I just mentioned like uh these, are all the available commands, like name version system pattern now, as you can see here, if I run the go command here, then the available entire list of the uh commands will be printed in the markdown format for the amoled. The output is bit different.

F

I can show here so this for a while this will this doesn't comply with comply with the uh present format of the documentation, because it's a bit different than what the cobra cla library generates it. So I have ignored it for a while, so I need to figure it out how to manipulate the output. So I kept it for kept it aside and focused on the markdown path.

F

Then. So, as you ask like it will generate the entire commands, like actually cpl machine ctl system, what are its um available, commands, sub commands, channel domains and type the whole car tree will be print, the documentation will be generated and then printed. So that's what it does.

F

This minimizes the effort actually.

J

Yes, yes, this looks really good. Actually, so because of the framework itself, we can generate usage and everything right.

K

J

Yeah so another question: right: we have the markdown output format, so the yaml format is for what purpose I mean. Do we use it as of now.

F

Yes, yes, we use it as of nowhere just show it.

F

Give cumulus so this is the uh yaml file which contains all the data like the name, description, usage of the command and the examples, so this is required for the generate for fetching the output. As you can see here, this would be entirely fetch from the this yaml file and then it would know, then it would replace this uh lines of code to the preferred one here in this case it's measuringctl.app, but uh here in this case it would just print the output what the command library is printed.

F

So if I just take the example of it, it's been a while. So if I just check with the called view, then, whatever the data given here for in this case, then the example would be printed as same as such, so it depends on what we give in the measuring ctl fence.

F

According to that, it will generate the output over here. That's what it does.

F

Near the queries.

H

I guess we are good, then.

I

Yeah, so silence equals consent for the other mystery ctl maintainers a couple of requests that I would have a detail as we push forward. So there's a couple of reasons why to go through this slight pain um and you in specific will I I I can't help but smile a little bit a little bit, but you and specific will well bear both the pain and the um and the fruit of this labor.

I

uh The pain because you have put into place previously well you've done a lot of work on measuring ctl, docs and trying to make them uh manageable, sustainable. There's a lot of commands. um Does anybody recall at the top of their head how many commands we have in measures? Gtl currently push wrote a blog post on this, and I did a quick line count of the number of lines in mesh for ctl.

I

If you're curious go to I'll, send a link or somebody will send a link to the post, uh but there's a lot of commands and aditya has done a lot of work here. He's become a jekyll whiz. Much to I don't know if he wanted to or not, but but I think at this point he is right, um so a couple of things um if the default template, so you haven't had a lot of time to look at this at dts.

I

So these are just comments not at you, but at the effort, which is we don't we don't use, underscores in the names of hyperlinks. That's a naughty thing. um That's something! Google doesn't like much. We use hyphens, just a super small request, the other one is. We would cut off the auto generated by on the bottom of each file.

I

We don't we don't need that, the other of which is the cleanliness by which this will actually not only um eliminate some inconsistencies in our what the difference between our docs lagging behind the actual functionality in this cli. This will help eliminate that, because these docs will be auto generated. So that's good, it's nice, that the markup is easy enough to style.

I

This will actually help us focus on like styling our docs, even better, because there's a bit of separation between some of the mark, the markup and how it's styled or the content itself, and how it's styled granted inherent to those three hashtags and to those three tick marks is a translation into html, but that's easily styled and so, um and we already have a lot of that style like so. To me, this seems like a big win. This seems like one of those frustrating things that it would have been so nice had we discovered this.

I

You know six months ago before we put in all the work earlier. What this will end up doing, though, is reducing it'll, make it more sustainable and bring higher quality.

I

It will reduce the duplicity, the the repeating of ourselves in our current docs, there's an uncomfortable amount of repeating ourselves in the meshry ctl reference guide, which, by the way, a dtf for most folks on here for some of the folks on here who might not be familiar with, like the actual reference guide, I'm putting that link into zoom right now, so everyone can go see what the docs are today from s3ctl.

I

In aditya, just to confirm like in your mind, the if, if you were done with what you were just showing in essence, the intention is that users or readers of the reference guide they may not even be able to tell a difference between what's currently published and the auto-generated stuff. Is that about right?.

F

Yep, that's right.

I

Any other comments for basically.

A

Yeah uh yeah, so basically, how would we see this on? The website would have to push this into a data file right.

G

F

For now, the ammo has been uh troubling me for a while, because the pattern of the uh yamala associate, as you can see here, is bit different than what we use for this machine ctrl docs section.

F

So that's the main problem and I've been trying to figure out how to uh manipulate the format of the yaml, and if we managed to do that, then it would. The work would be much easier.

F

That's what I'm looking into for a while, but.

A

But if we can't do it, then we can just manipulate it on the website right. We could do it either way.

F

Yeah, we can do it already. There is also an option like there. They also they also give any reference to it like, um instead of like customizing the default output for the command generated by the cli library itself. That is an option to generate the custom tree or the justice command alone, for uh in this case, for the markdown one. I've used the three functionalities such that. That's the reason why the entire section of the measuring cpl markdown files have been generated else.

F

If I just give this markdown custom, then it would generate for the uh single command alone here, in this case, it's just true that you would generate for only the message: ctl command alone. So if I just give like generate markdown three custom, then it the command would know. Okay, uh I need to generate a markdown files for the entire tree of the measuring ctl, so measuring ctl measuring secure system hdc. This is the ctl system channel the entire tree will be unknown and then printed so for the yaml one.

F

I will look on it and work on it for a bit.

G

All right sounds good. Okay,.

A

Any other questions guys.

G

F

G

A

uh Thank you that was great moving on, uh we have pranav pranam. Would you like to demo the progress.

F

Yeah I'll share my screen.

H

I

I'm so like basically, oh nice, I think you're about to do it. I was just gonna ask if you would give a little bit of an intro as to what it is you're going to talk about.

E

Yeah, so I would be talking about how like basically improving the ui of notification center basic notification basically gives us the information about the status of the operation, which we apply on the adapters which are connected to the mainstream. So my work was basically improving the error messages which comes from the failing operations. So right now I focus on easter adapter and its failing, install operation. So earlier there was that there was the information which was coming from easter.

E

Adapter was very limited, as in like only the summary and the details of the failed operations was coming on. So what I did was I enhanced the event structure which brings the messages from the adapters so because of which now we get the cause of the error, error code and remedy of how this particular error can be resolved.

E

So this was my approach and now I'll show you like how the new enhanced errors look like.

E

Like if I want to install each service mesh.

E

Then it should give us a failing operation message.

G

E

Yeah, okay, so this is the updated ui which I and which I added so this information basically coming from the adapter itself, but these two fields are empty just because the error dot go file of each adapter doesn't include this particular information for this operation. So once we update that, then it would be like it would be coming from there, and I also added the error code, which was from this, um which is related to this failed operation. So this was my approach.

E

A

I

You've got a winner, I guess, I think yeah. I think, there's probably a number of people who are holding back some comments um in part, because the outcome of what you're showing pranav is like is very much needed and has been something that we've long talked about enhancing and that there's a lot of people who put in work into trying to bring a uniform approach to capturing errors and then going the ex like not just an extra mile but like a couple.

I

A couple of extra miles to you know: capture errors, provide them with a unique identifier, kind of classify them, then go and define, or go and suggest like what is potentially happening within the system. What the probable cause for the error is and then even taking it a step further and saying or suggesting what might be remediations that could be taken. You know, steps that could be taken to fix the issue and there's a lot of that data floating around and it's landed.

I

Actually, if this is this will answer one of my questions. A lot of that troubleshooting information has landed inside of an error code guide inside of mesh redox, and if you do me this favor, if you were to click inside of measure, ui on that notification, there's it looks like the error code is hyperlinked.

I

Does that hyperlink land us on uh the error code guide.

E

Yes, it does that I can show you one minute, like I've added the link for their records idea.

I

Cool okay, um not good enough or or rather that's fantastic, that's highly appropriate. You can. I will challenge us to take it. One step further is shivam.

I

Shivam y on the call who's, who is the contributor that was working on enhancing the error codes guide? It just got.

I

No well um one of the last pieces of feedback for him had been to ensure to to make a couple of changes. Can you open up the guide for everybody else? That uh may not looks like we're.

K

I

We're talking all about uh the docs today and how important they are, so the error code, reference guide, this catalogs every known error code that we capture within any measury component. So any component, that's using the mesh kit, library and it's a library that our projects use to to define and capture document errors in a consistent way. Can you do me a favor and click on any of those rows.

I

And then click on a couple of others like okay, good, okay, the probable the probable cause on that one is.

I

Angel is to go look at some docs and get up for rest period.

I

That's probably suggested okay, um but okay, so the last.

E

Piece of feedback, I would just think that I think two information is not written on the files because of which it's not shown here.

I

um Yeah, actually, I hate to tell you I hate to be the bearer of bad news. The the information isn't there, because the people, the contributors who created the errors, didn't take the time to fill in probable, cause or suggested remediation.

I

So it's just missing in the code, but that said, there's a good framework and some of these are filled in more of them need to be filled in so for those that are looking to get into golang, who are on the call or, if you're, already into goaling a great place to make an impact is in in in these errors and documenting these errors in in golem.

I

Anyway, the point was going to be is that you should be able to? There should be a unique anchor, a unique a tag for that is a combination of the mnemonic in the left-hand column, combined with the identifier in the second column. If you were to add that as a hashtag as an anchor in your query string, you would then be able to land the user right on the specific error and potentially even we could have a modal invoked.

I

That, like opens that particular error, or if we publish those to individual pages, you could direct to that specific page.

G

Right now I didn't.

I

Think that there.

E

Is um I think that yeah? I was also about to say that in the dock itself the anchor is not added.

I

Yeah I realize that that's exactly why I'm asking for where shubham is because I'm going to reinforce and hammer it down on him that he should do that and then uplift him and support him in doing that. um He's not going to get it done.

I

So uh there's an opportunity for anyone on the call to do the thing that I've been suggesting, which is to combine those two fields into one: don't don't don't have the columns separate or if you do make sure that there's an error, an anchor tag that has both of them combined, so that you can easily direct people to the specific error code and then in meshkit whenever there's an error, that's generated whenever someone creates an error, um the message that's given back sometimes include like if, particularly for mesh rectl, I think we'll give back a url to people, sometimes we'll say: they'll execute a command, and sometimes we'll say.

I

Oh, you know this. This happened. Here's the error code and here's the hyperlink well that hyperlink right there. It too could also have a direct link to this particular page so um anyway, back to what pronoun is doing, but all that that's um fanta, the thing that you're doing is um fantastic like we really need enhanced um notification center um data about what happened. It's equally. So, okay, that's good! It's important to for us to provide troubleshooting information in the event of something negative happening, but it's the same.

I

If you go back to your code on the flip side in the event that something positive happens like oh istio was provisioned and there were four control, plane, components and blah blah and, like you know, he hears the things that were like it's. I'm sorry. If you go back to your um the measuring ui and yeah yeah, if you go back to the uh not in the code prado but uh okay, ui.

E

I

Yeah and you go back to your example, where you click on the badge on a notification bell- and you show the details here if this was a positive result like yep istio provisioned.

I

Well, the details of okay well like where or uh uh in what fashion. Or did it do like, because people can ask for mescheri to provision any of these service meshes in a fairly sophistic, um highly detailed way, and so were those particular configurations that they'd asked for in fact applied. And what does that look like? What's the summary of that?

I

So as you and a collection of others on the call, abi and and others who are looking at this messaging and notification, this messaging framework and notification center, um I think it's easy to myopically focus on the use case of a uniform framework for capturing and cataloging errors, but that same messaging framework. We want to capture positive messages as well and the details that we might want to show a user.

I

A

I

Now, to help refresh everyone kind of in this area, um the way in which you facilitated the code that you've changed some of what you changed was a protobuf file. There's each of the mastery components speak to one another over grpc part of grpc.

I

We use protozoa.

I

And the particular one that was created like three years ago is called well there. I think there were a couple anyway. This was called mesh ops.

E

um This is the one which is my shops. I think.

I

Okay, did you make changes to meshes.proto as well or not.

E

No, actually, I think that only this one is used. The other ones are not used right now, because this one includes everything which is written inside this, and this I feel like that.

E

We are not using this one and this one, because the output files are just for the mesh ops, one that this one we are using actually and it's produced out of this, so I feel we are just using the mesh option.

I

Can anyone else have a call confirm that.

I

And ashish might not be with us or able to answer so. Okay, uh so of what you had added pranav was this: was it? Was it the? It was just a couple of additions to an existing message.

E

I

E

This message event response: I added these three fields. These were already there, these ones. I added these three and then basically like I can show it over here. These fields gives come, brings the message from this particular part. So when we are installing it, so it's first time to it gets trying to install the html service mesh and then, if it fails, then it takes this information from here and then streams it to the machine server.

E

So I added these three functions from the mesh kit, like I'm using mesh functions to get the error code, remedial calls from for this particular istio, adapter and and then I'm then streaming it to the.

A

A

I

Krishna, I muted you just in case you so pranav. um What do you suggest that next steps here are.

E

um I feel that first of all, like this same even structure, is also used for the like successful messages. It's the same structure if it gets successful, so it it just take the two information that is summary and details and like if it fails, then you take the extra information which I added. So I wanted to ask like we would. We will be using the same event structure for both successful and failure messages, or we will be in different structures because for failure we might need different fields and first um for successful messages.

E

We might need different fields, so I just wanted to ask them, but depending on that, we can iterate over other things. uh So uh so now, uh let me confirm, like please confirm me that success and failures are coming from the same functions right, so they are streamed from the same uh so probably uh success. uh Full events won't have any error codes or we we should provide actually uh one more uh struct there, that what is the type so type can be one.

E

You know error success and all so. The second approach looks better to me, since it is, you know, scalable as well as lesser prone, but maybe yeah, just adding one more, uh even adding one more structure like just defines like. If it is, you know, successful, uh snack, bar or error, but it's like that, because before my change, like every adapter uses this same structure only for both failure, as well as successful machines.

E

So if I do make a change, then I have to make change like among everything like every code which uses that substitution will that be like fine, but right now like what I'm doing. Is that I'm not showing the other fields on the ui? If the message is not error like if the message is not the failed one so because of which I can use the same structure for both failure, as well as the successful messages so like, which one would be the better which, if someone.

I

We can yeah yeah, that's, that is the question, um and so actually we won't be able to answer it on the call, because it does it'll require writing down a couple of use cases. um The I just sent a link in zoom chat to this particular design. Spec- and um this is this- is where we need to ask that ask and answer that question uh yeah yeah. We just need to press press in here.

E

Other than that, uh since you have worked so hard in it, uh so basically, uh let's let you learn another separate poll uh discussing.

F

E

How success things can be done? Okay, yeah, okay, thanks.

I

Any other comments um uh abby when you, if you do schedule a call um you'll just for those that are interested, you'll post, that in like either in slack or on the discus forum, so others can join and kind of um iterate. It might you know just as like, as we move forward into v0.7 like into the next release, that um this is a great example of taking there's a number of design specs that we need to go under review and kind of, so it sounds like we've already got.

I

We might have a topic all lined up for next wednesday. Next next measuring meeting, which would be you know, maybe it's a completion of this design spec or maybe it's just an iteration like a step, four, a clarification of some of the use cases and how closely they're cut okay. All right, that's great guys. This is really exciting.

I

So yeah, uh please do you know, please grab a hold of abi or pranav. If you're gonna dig into that area, it does. It is a pervasive thing: it does touch um all the systems. All the systems should be hooking into should be sending the same using the same message format so that those can be exposed to different clients and in different ways.

A

All right thanks, pranav uh moving on uh push, would you like to talk about the portfolio.

A

A

So this is the new feature in our system: dashboard command- that is port forwarding and, as you can see, it ran successfully, but it started a stream of errors which is not like outputted by message. Ctl, it's from uh client co of cuba needs and the command actually ran for almost four hours. So uh the problem with the actual port forwarding that it doesn't work for longer time has been dissolved in our case, I'll just run right now to like show you guys.

A

Let me first clear it.

A

Okay, so the good thing that was that is now it. It runs on a random port which is available.

A

Previously it was hardcoded to 9081., which is like the standard visually imaginary port, but now, like uh in case of port forwarding, it will like search for a port with which is available and start like port forwarding to that port.

A

A

So yeah, I'm logged in things are working.

I

It might help provide clarification to the user if, if you end up showing the source, port and the destination port, or if you end up uh to helping clarify a little bit about what the local port is, that's being forwarded to the remote port and then.

A

F

A

I'll add and log log for that, like um eight zero, zero is being forwarded to three zero. Three, like the the sport, yeah I'll, add a log here.

I

This could come in really handy so just to for recap for everyone else when you run messages just to clarify the behavior between measuring ctl system dashboard and mesh free ctl system, dashboard port forward is that.

A

Okay, so the difference between this is like we are actually running the actual cube, ctl portfolio command with this flag, but with dashboard we try to get the accessible measuring. Endpoint like we are. We use uh load balancer type service for machine, so we try to get that accessible endpoint uh with dashboard using our mesh kit functions. So if I run just dashboard.

A

It's trying to get the service and yeah so.

A

This is the url like this is where mini cube is running, and this is the port where, like mini cube, is exposing that ip of the measuring pod. So that's how I'm accessing my shirt here? Okay, so that's the difference between normal dashboard and the port. Forwarding port forwarding will come handy in any cases where cluster is does not support load balancer.

A

It doesn't have load balancer service like exporting to the external ip and if.

A

If it is also not like it is not mini, cube and mini cube actually exposes its network on the local host like like this, but other cluster like kind kind, doesn't do it, and other tools also does not do that, so it will come in handy if someone is running on those kind of clusters.

A

I

Do you think you worked out all the kinks?

I

uh What did you so last time? I think you had said that we talked about this. You said that uh portforwarding doesn't really seem to be working like you know. Sometimes it does. Do you think it's solid at this point or.

A

It is solid, but uh people might like it. It is uh what should I say it is solid, but it will start streams. The like error streams, like just I was showing you before mohammed. I cleared my terminal yeah, so it will start streams if it is being used for a longer time like longer than five minutes ten minutes, then it might start it, and it is not like a guarantee that it will.

A

It will definitely start those error streams, but it might start to better stream and that those errors won't explain what is going on and just to like, if, if it will still work, even if those errors are being printed in your console, it will still work, but it might also not in some cases.

I

Should those are those things that we like yeah, you know think about that a little bit. Maybe those are things that we should show to the user or shouldn't or when we do set an expectation is to what that means to them.

I

Maybe not a high priority thing that.

A

I, like I tried to hide them, but there was no such option provided by client. Go to hide them like clientro is also using some other package inside it to like. uh Do the dialer thing dialog between the pod and here, and it doesn't verbose it like it doesn't, so it will keep printing them and some, like some errors, are just normal. You can still work with your liking.

A

Ui, your machine ui will still work and even if those errors are being printed, but sometimes it might stop working, then you'll have to just control c and restart uh preferred that's the way to go.

I

I I dtt too, anybody else have feedback for push.

F

And yeah push um in the recent review I gave on the pr like I have showed like for ports running already on nine zero, eight one. The error message has been printed and you already saw that right so on such cases uh how the user will be notified. If such port all rises, the user might not not know right it. Would he he or she would consider like this- is an error message uh what's going on why this port is already assigned in such cases how it can be handled.

A

Okay, yeah, okay, it is not here, let's go down back there, okay, so previously what I was doing here, I was just slightly using 908 one just like this: uh it was kind of hardcoded and then there is a function that I like we added here. It is a get a female report. What it will do is it will actually go. Try hit random ports and check if it is available.

A

Okay, so if it is available that port will now be used so like, let's, you can see here like previously, the port was three four three zero one, and now it used four six, six, three five, so the case of like the port binding and bind it to some other service is like not. It won't happen now.

F

All right glad to know that.

A

Any further comments, guys.

A

Okay, I guess not thanks viewers uh moving on avi. Would you like to speak to your topics here.

I

E

Okay, so uh basically uh yeah, so actually uh there is a basic thumb rule of using uh hooks, that is, it should be used only in the functional react components. So uh so actually the error were all originating from the usage of hook uh from the functional javascript. uh You know uh components uh that are not react once. So. That's why uh we were. You were getting the dashes in the ui uh while opening our configurators uh some items in the configurators, so so that is the one fix other than that.

E

uh Just just realize that uh we have, we are generating. Some uh schemas, like uh dynamically from you know, from the adapter adapters are uh themselves generating. Some schemas for us are to be rendered, as are just forms so from there uh we have, we have been. You know, uh we don't know about that. uh That accumulates interesting. Can it can be a valid feel actually because rjs have always always uh accepts a type field, but when you are using x cuber it is interesting.

E

So uh you you know, the schema won't come won't be coming up with any type of because that is uh already already understood from the key itself. So that's why uh there is just a tale from that. We are uh that uh adding support for kubernetes interesting.

E

So these are these two are the some exceptional fields you you're seeing these two are some exceptions that doesn't contain types and we were just missing on those things uh due to which we are running into some ui problems uh schema our problems, so these two are the fixes right now, uh other than that. I was also working on some life cycle, testings, uh so cyprus, but uh it is quite not uh very ready to demo. So these are my updates today.

I

Abby there are a couple of projects that um that I'll I'll point out, as potentially having some reusable tooling here in terms of validation, schema validation and.

I

This is this is related to this is related to, but not directly, aimed at the issue with kuma. I expect.

E

Yeah so uh so puma they generate our components dynamically once like ashish. uh As said in one of the previous calls that what kumar does is they generate complements after are they running inside the kubernetes cluster? So they don't generate static components and we are, we are, are totally relying on the static component generation from all the adapters. So that's why kuma is one of the uh exception, uh so we have to think something uh else for handling the kumar components. uh You know the generating the schematics for kuma companies.

I

The validation that you do so inclusive to this pull request. Is there a validation in there validation of the schema.

E

Validation of uh validation, of which schema actually uh by validation, davin.

I

uh That that uh any of the objects that we're going to present uh have the required fields like.

I

Validation, yes,.

E

Yes, uh okay, okay, okay, so that is, that is validation, uh so this doesn't cover that validation actually. But what says? uh No! No, since there are two three, only two three four fields like the name title uh as well as type and like these three fields, only that that should be always present inside us a schema object for rendering it into our gesture. So actually uh this looks a good idea to uh create a function that you know uh well gets all the schema at the first place.

E

So probably uh I will create a video for it, or I think nitish has already done some working in the past. All right so probably try to extend it.

E

Yeah, so this doesn't have to do anything right now with the validation. It's just about catching the new case. uh We just get informed about so yeah. That is, these are the updates actually other than that the kuma components are. We are not hitting into ui crashes right now, uh because because that is just got here, yeah, you know in this this portion, uh but still uh it's not ready.

E

You know, uh because because we don't have any schema, so you can't actually use my shops v2 for and generating any you know using kuma itself, but everything else it would work as fine as it was before.

E

Yeah any other questions.

I

Oh yeah, um we just have a couple of minutes left. There was a pr that was merged um during today's call and if that pr affects the build checks that run for all of mescheri's, pull requests.

I

Mario, if you don't mind, can you just let everyone know what those are just the top level checks that are going to be run now.

K

Hello yeah sure, okay,.

I

90 more seconds.

K

I

K

Let me start sharing: okay, just a sec.

K

Okay, so uh can you guys see the screen.

K

Yeah awesome all right uh yeah. Well, I should be over here. Support requests just close that screen. Sorry.

K

Too soon, okay, sure, um so basically the this pull request is um it contains some tests. um I don't have uh measuring running, uh but basically it's for the a configuration, um a route pages, and so I had to comment out and so I'll just show the the file change itself.

K

uh So, basically, um first of all for the workflow, I added this uh instruction to run measuring istio, along with a ci, yaml workflow and just a small note added a on the mbm usage, just a recommendation on the ui readme and the most important thing is uh the data cy attributes that were added, and this helped uh like for the filters header, a measuring, app pattern and um yeah. So these pages and the actual test is um it's basically, um okay.

K

So here are the helpers, but basically uh sorry, sorry, I'm a bit uh rusty with uh demoing this. So this is this started with since uh december, but uh yeah um without further ado. So basically uh here are the tests and right now this store commented out because there's still some issues going on. um But um this regards further investigation of uh why this uh um a kubernetes objects are not a found during a like a when applying the or deploying the the configuration a, but the filter works correct.

K

So this array of objects is passed to the to this template. uh So this cypress test it'll just go through a certain steps to a just um intercept some requests, and this is running both the measuring server and measure ui, and once that's done, uh the converter configuration page is uh visited.

K

uh We wait for uh the request for the items from the from the grid to load and then uh there's just this test uh template that make sure to um upload a file. So the data is taken from a test file passed a through the like test data object and then so we wait for the a upload to finish and once that's done um so that request is intercepted. And you know all of this.

K

um uh A code is just so that that further actions just wait for the for this promise to resolve, and then you know we can continue with the uh actual deployment, and so once uh there are some uh checks, but so now we're just looking at the filtering uh but eventually uh and further appears. This should also work for deploying.

K

So this deploys now uh a filter through a local file upload uh through the page, and it checks that a the response was a success response and also that the body structure that it's as expected in and um well yeah. I think that's it not sure. If there's any questions.

K

Oh and yes just one more thing, sorry so right now, this is the um the filter being applied. uh This is taken from the samples, so it's uh the http of um envoy from the wason filters, um a samples, uh so I'm not too familiar, I'm not familiar with uh webassembly. So I couldn't. I would not be able to give like a great explanation, but basically this is what it's used to test. It looks it's looking good for now.

K

Any comments or questions um silence.

I

Yep, no, you just wanted everyone to be.

K

I

Just wanted everyone to be aware of the fact that there's gonna be new tests and to look out for them if, if there were I'm assuming that they are all passing as of this moment,.

K

Well, I want for the filter uh so as we.

I

Mentioned the other day, yeah, let's comment it out, though right so all tests are all merged. Tests are currently passing. Aha.

K

Yeah, so so the filter one is passing and the other ones like there's still some things going on, but uh once that these issues are fixed, um oh, we could just uncomment them and it should be passing as well.

K

So so this is the location where this test is so uh not sure if um ashish or I don't know who who wanted to take a look into that. um But maybe we can take that offline through slack.

K

That's all thank you.

A

All right thanks thanks mario, so since we are over time, we can discuss the rest of the topics on the next week's call. So thank you guys all for joining I'll, see you guys next week at the same time, and also on the that's also in the community call on friday. So thank you all for joining bye, bye, bye,.

B

E