Layer5 Meshery Development, 9 Feb 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Meshery Development Meeting (February 9th, 2022)

Description

Meshery Development Meeting - February 9th, 2022

Agenda:
- Discussion: Context parameter to control service type to be deployed (#2968)
- Improving navigability of error code guide #4981
- New UI changes for Sidebar #4926 #5042
-Test coverage in Meshery, Build and Release Meeting
- Kubernetes Icon not visible on pattern configurator
- Sample feature clip (sequence 1)
- The service mesh playground (ux, spec)

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

All right and we're live.

A

Well, it's wednesday february 9th. I think I'm your host today! My name is lee. Okay, no most folks are on the call, if you just joined um then be sure to crack open the meeting minutes I'll, put a link into the zoom chat for the third time in case you missed it the other two times, uh but we've got yash subum pyush, davindu, manov, garov, aditya steven uh me good, uh there's a couple of folks that sent regrets today.

A

uh I'm going to take a quick note.

A

Does anybody have topics to add to today's agenda.

A

If you do drop them in, we probably have space, but if not, then we'll um continue. So this is the mesh redevelopment call it's held every wednesday uh first topic up is.

A

Sure, okay, so this one this, this is an open issue by.

A

Opened by abhishek, what is this on? Is anybody on the call um familiar with this issue? Is this something anyone else is working on.

B

uh I think it is about uh letting the users configure the service type, whether they want not bought or load balancer, and whether we should add that to the mesh config file.

A

God that makes sense is anyone. Is everyone in favor of.

A

That does anybody know of other uh projects that that dues excuse me that do such a thing.

A

The the common way that that some of the projects might do that would be through like a values, file and helm if they have a helm chart. Are there other ones that offer this choice.

A

I wonder if we're taking it too far, I don't know if we're trying to make things too convenient too many options. If we should just expect that users of meshri are savvy enough with kubernetes that they would configure their deployment as needed, so you should think you've been entertaining the concept of a dashboard command. I think we've had that concept for a long time. um The dashboard command would be to help would basically be a small proxy that would export mesherie's ui uh would well would proxy it local to the user's desktop.

C

And that's how it doesn't work. We have like a like what we basically do is we try to like. We know that measure is using load balancer, so we try to grab that ip the address to it, the accessible address and use that, as like the access point for measuring ui, we are not like. I tried to like implement a implement, a process which had opening a proxy or port forwarding it, but they were not good.

C

They, like those methods, were not good. um So that's why I changed it like used whatever we already had in the mesh kit. Would you recall what method you used?

C

I tried port forwarding and I also tried to uh like create it start a tunnel and then grab the ip, the accessible ip of the like a measuring ui.

A

With the port forwarding, were you leveraging um kubernetes go client right? Okay, what what ended up being the challenge there.

C

It was like, uh like I tried to use, uh I like I implemented similarly to link id. What linkard is doing. Linker d is all support forwarding their ui. I did the similar implementation.

C

What was happening was uh think of it like this, that, after some time like that command is running constantly in a terminal it becomes stale, it stops working.

C

It throws kinds of many errors. I don't I don't remember now, because it's it has been like too long, but it it was throwing errors. After a certain time.

C

And that could be five minutes, maybe just after five seconds it was totally random and it was not a good idea to use that.

A

Had you yep, had you run link your ds command and I had seen similar behavior like. I wonder if that's a commonplace issue.

C

uh I didn't tried linkard's uh like linkadie's dashboard command. Did you take it.

A

As you reviewed their code, did you take a similar approach.

C

Yeah almost similar approach, okay,.

A

Is there a, can you guys refresh me, is there a cube, ctl, dashboard command.

A

Or like a plug-in command, I thought there was.

C

I'm not sure I'll check.

A

Because it it might be uh well, that would be another example, because I just wonder like that seems.

A

Like our ideal methodology, because, if you're doing poor forwarding you're basically circumventing any concern about what ip address it is or what type of networking it's doing, it's just you're you're you're, creating the networking you're establishing a direct connection between the user's system and uh the pod mesh reserver pod, uh and so that's ideal. So command line cube proxy. So another thing that we could do is at minimum. We could document like within the mastery ctl system, dashboard command, as we attempt to create that proxy and have that command work.

A

One thing that we could point out to users is, they can use, they themselves can just run cube, ctl proxy um and then use cube ctl directly to proxy mesher's ui.

A

That's something we can do.

A

Nobody's familiar with whether or not there's a cube, cto dashboard command.

C

I didn't find any cues hit your dashboard on it.

A

Maybe I'm thinking of like an extension or something plug-in.

D

A

Else is I'm going to see more.

A

I can't get my terminal.

A

uh Your voice is pretty low. What did you say? Hello, hello, I'm audible, yeah, it's still about the same.

B

Now ashish mentioned that there is a plugin, but he isn't sure what it does.

A

Okay, well, all right, so so I guess back back to so the reason we're talking about mastery ctl system dashboard is because there's an open request to enhance the intelligence of mastery ctl system start. When you run mastery ctl system start, it will say it will attempt to figure out the um how you as a you, what url you should use as a user to connect to mastery's ui.

A

And occasionally it has problems doing so depending upon the type of kubernetes deployment. That's there different kubernetes deployments will support load balancer as a network type or or other types of networking.

A

Let's do okay so, and then I was suggesting uh this really relates like one of the ways to circle to overcome this issue, as opposed to supporting more configuration items for users is to just have a dashboard command.

A

So that's why we're talking about dashboard? um Who else has.

B

Thoughts on this uh another another point here like is that the measure since misery can also deploy be deployed outside the cluster. uh It doesn't make sense to have this configuration as uh as a configuration for measuring as a whole, so and deploying machinery outside the cluster is a valid use case. I guess so.

A

So if there was a parameter added to mesh config, it would need to be platform specific.

B

Yep yep, okay and we don't have any platform specific configurations now.

A

And so you you're opposed to create, I mean okay, so say more, do you think creating platform-specific options is just kind of another minus one against.

B

A

B

uh Yeah and like uh I think we had, we have had discussions on this like uh we should just let the power users modify the docker compose file or the kubernetes manifest files uh to tailor it to their specific needs, which we can, which we support. Actually, so, if the user actually modifies the manifest file, we we use those manifest file instead of pulling pulling the pulling our default manifest file from github, so they do have an option to uh make changes to the default, manifest files or the helm charts yeah.

B

Do we have that documented? I wonder I'm not sure I'll check it out.

A

And I don't know, I'm just wondering aloud if this is a good step to take or not I'm not. I do know that it's a hindrance to or kubernetes networking like if mario is on the call he isn't, but he stumbled all over it uh couldn't couldn't get past. It needed a bunch of help to figure out like how kubernetes, how do you expose a service in kubernetes.

C

I also think that we are like over complexing. It.

A

You're you're learning that um well.

C

Like it's just giving too much uh options to normal people like new people who are going to use measuring.

A

Yeah, it's funny because that's actually the argument for this is to say like oh, maybe our current configuration our default configuration works for them. Maybe it doesn't um if it doesn't and they're new to kubernetes they'll be up a creek and they'll, be um whereas yeah I don't know so I think it so I think it so. Okay, then, what we're saying is like a solution is to expose a different configuration option. It's like well, you probably won't figure that out either.

A

Maybe what they would figure out- or I would think would be much more likely to figure out- is, if there's a dashboard command that they might like. Quite often, they would probably stumble across that and figure that one out and the dashboard command would actually figure it out for them and create a proxy which I don't know like yeah there's, there's a nice.

A

The nice asp, like the the issue in general, I think with the dashboard command is well, it's kind of what push was just saying which is like well, it's kind of a hack or sort of like, maybe you wouldn't say a hack, but you'd say it's. Certainly, a temporary.

B

Yes, temporary solution.

D

C

I also read on some document that put forwarding for a long time is not good. It's bad. That's. I also took a step back on that uh that idea.

A

Sure yeah as a general statement, there would be a lot of argument with it being bad because, like that's, how we're having this zoom chat right now is port forwarding like this, like forwarding, goes on like all the time, but um but I think point taken in this context that, like it's a it's, it's there's a performance hit for the most part. In this case, that's not really a concern, but there is. It is a transitory thing like it is.

A

This is a temporary thing where someone would initiate a proxy and for purposes of a user deploying measure and trying to locally connect.

A

That's that's probably okay like it, and they know that if, if they execute that command- and it's not backgrounded that um there's a foreground ongoing, you know task running in their terminal if they kill the terminal, they're, probably going to lose their connection like if this isn't, um if they've seen this kind of a thing before anyway, I guess- I guess my and then okay, so I guess the point of saying this is that the the dashboard command might be the right thing to lean into because people are.

A

Newcomers are more likely to discover that and find it valuable and easy to use because it just figures it out for them they don't have to go learn what service type works in their kubernetes environment, while people who are deploying meshi and running it long term can do understand what type of networking works in their environment and ultimately can go configure that um so yeah there's there's a few different answers and I don't know that we have any of them down.

A

Pat one is expose a service type, and if it's exposed it doesn't it doesn't mean that it has to be in there by default, so we don't have to over complex the mesh config by default. If that doesn't you know necessarily make sense, it can be by default. Its load, balancer, is what we've chosen, but if someone wants to overwrite like go, looks in the documentation, they can either supply their own manifest use a different, install method other than mesh for ctl system start.

A

um They can then add, add this option. um We could also have the dashboard command. um We can also just have documentation that says: use cube proxy as another. A fifth solution, so actually most of the time starting with docs, is well conceptually. It's the quickest thing to do, um and we certainly want to have good docs that explain to people that they need to expose their service.

A

I think, if p, you should think I think we were you know we were thinking like the dashboard command was slated. It was something that we wanted to do. It was kind of it was coming later in the road map and we're we're almost to that later date in the road map. Now.

A

Does anyone else want to characterize like who else has anybody else have strong feelings about whether or not to go go forward with this kind of thing or not?.

A

If there was a platform specific service type, um would there that be an indentation under platform or.

A

Yeah, are we just spending time on stuff? We should be spending our time somewhere else.

B

Yeah, I will open an issue and, like I will document the fact that they can override the default uh service.

C

A

To navindu's point that is more or less the same thing as being able to define a service type in mesh config, you know more or less, which is good right so, which is to say that like well, that's to justify that there really isn't too much value in being able to specify it in mesh config if you're, basically able to specify it in a different file.

A

I think this whole thing is quite debatable. You know any which way, but that's.

B

And in case of the dashboard commands, maybe like we can, we can check how the other other service measures or other kubernetes uh applications are using that like uh exposing their service. uh Maybe as soon as mentioned, like linkedin, has a dashboard command, and maybe we can dive a bit deeper into how they are achieving that and okay.

C

So one thing that I want to like to say again and that is that, like port forwarding, why did I step back on? That idea was like if someone is using measuring ui, he is going to use it for a good amount of time, not just uh a minute or two minutes or five minutes. At least I think more than 10 minutes, okay, and within that span of time the port forwarding is going to lose the connection.

C

It will start to throw errors and, while you're working you will lose the connection because of that and things will stop. That would be a bad bad experience.

C

That's why I declined what linkardi was doing in their dashboard.

E

Oh not necessarily important, but which, which types? What does it mean series types which types of services? What are the service types that we have.

A

Load balancer cluster, ip and node port.

A

Kubernetes network.

E

Yeah we do need to have parameters and even more than those or I mean as a number, even more so.

E

As an answer to your question, if someone has another idea, it's against or pro I'm against complications, but I think on this one, that's why I asked about the service tech.

A

It makes sense and yeah they are supported. All three are supported. um What we're talking about is, like the user experience around the way in which people configure those.

A

Yeah, um so too is uh aditya's. Point like mini cube is probably another an example of another system that proxies its dashboard. I think istio has one um like th that doesn't forego the comments that push that you're making it's just if we're having failure, if it kind of hiccups over time. It's like ah check your code.

A

There's uh there's you got, you got some, you got some port leaking, some memory leaking or I don't or I don't know what's going on, but I was like.

D

It shouldn't be happening.

A

I think we've beaten that uh any other comments on this one.

A

Schubert, do you want to talk about the error code guide.

F

G

uh Yeah so uh so the issue was to increase the navigability of the error code table, so we had like uh three more columns, apart from it like showing from the earliest, which was probable, cause, suggested remediation and long description.

G

So uh what I did was uh I added a clicking behavior in which, if we click that uh click that row so user can see. The extra information that that particular error, name and error code has and uh the other thing, which was that that earlier error name and error code was in the different columns, so so that the user, so that the google could find it easily have made them in the single column.

G

That's what I have done on this.

A

That that is the moment of feedback. Folks, who's got feedback.

H

uh Can you show the one which has the probable cause or pressure remediation one like? Can you show how it displays yeah yeah, so which one which was yeah, no yeah, that one okay.

I

I just want yeah.

H

uh Can't we, you know, spread out the suggested remediation part a little bit like it's on. It's I think there I don't think we can, but it's looking a bit cluster, the suggested remediation part in the middle.

G

Yeah we could do that.

J

So probably suggested remediation will be a short descriptions.

H

Yeah yeah, that's yeah. That was the only issue I found. Otherwise, it's looking.

J

Good, uh I was telling that suggested remediation will uh be always uh will having a small string, so it won't be any problem by the way. It is looking fine to me.

K

Just one thing we can add is uh on hover. We can show an uh drop down icon, so a person would know that it expands.

G

Yeah yeah, so, though, only those uh error row, which has the uh information, so we could have over, like your drop down icon.

F

Anybody else have feedback.

C

J

C

uh What we have this suggestion recommendation under probable causes only there could be area of items instead of doing this. It should be if it is a dropdown sort of card popping up.

G

Like array of items like uh we have like more.

J

Probable cause, we could have suggestions.

G

J

And below suggested, we could have long description like probably cause poland, its name. Then the suggested drama edition colon details, long discussion.

C

Poland, details.

G

Yeah yeah, I understood so yeah. That's also is an option uh so like what do other things about it, that about this, that provide will cause just reading remediation and long description like one with one below others, one below the other.

F

G

Yeah yeah so yeah.

C

This way, the card also becomes a little shorter and.

C

And this suggested thing like it's looking squeezed quite much.

G

H

G

H

Yeah first, like long description, then private cards then suggested remediation. That order would be good. I guess.

A

Nice uh good yeah, that's some great feedback by the way, so the yep, the long description should come first, um because, if you're looking at what the probable cause is and suggest and then how to fix it, and then you read what the description is. It's like! Well that probably want to know what it is and then the other thing is um much to obvious dismay like no.

A

The suggested remediation should in fact be a little more verbose than we're generally making them what we have are we have our contributors, who aren't necessarily putting in a lot of time to fully describe what someone should do to fix something. So here's one like this suggested remediation on 1019 suggested remediation check. If your request has valid ohm config well, I have no idea what that means or how to like, where.

A

Yeah, I don't know what we would do right, there's a by the way, there's a typo there and a phrase in the long description phrasing, uh but yeah. So we we by the way we may have multiple causes like um hey. If you do measure, ctl systems start and it comes back with uh whoops, couldn't connect.

A

Well, why couldn't measure cto connect to the? Maybe your network went down. Maybe you have an invalid certificate, maybe you're above, like maybe 15 different things like really really 15 or 20 different things. What should you do to fix those things? Well, probably some different, some of the same steps for some of the same one different ones, but some different steps for other ones. So, like actually there's an issue here, so mono set a keyword about an array.

A

There's an open specification, design spec for a messaging format, a messaging framework in which these there's the error code system- that's being used, it has it doesn't it itself doesn't have a set of arrays set up like we have unstructured data.

A

That's happening here where all we've really allotted ourself is a single is like is a string for probable cause, a string for suggested remediation and uh for long description as well, there's no room for markup in there by the way, and there's also no room for a structured way of specifying that there are multiple of these.

A

So that's in the framework itself. The error code framework that we have it could be enhanced to have to separate that out, so that we can present. So the presentation of the information can be more easily handled, but in lieu of that I mean- or rather despite that, we can still.

A

We still want to plan for not like the the alert column is well. It is formatted it's wide enough to fit the the severity of the alert it's not really intended, or wasn't at least in this design intended for a sentence or two about suggested remediation, like yeah when you reformat. Basically, you need to reformat and have something other than just a table row within there. You might put a d, you might do a call span of four one, two three four and then just put a div inside and then manipulate it from there.

A

You might. I think that the drop down table is quite convenient and is presents people information right in context of where they are like the hail mary here, the thing you could do, if you can't figure anything else out, would be on row.

A

Click is to pop up a modal, and then you've got free space to do whatever you need to do it's not as ideal it's a little bit sort of I mean it's it'll be fine, but I think you can still work within the confines of what you have here to arrive at um at at a place where some of it it's some of it, isn't quite um squished yeah. If you go instead of going horizontal, if you go vertical with you,.

A

But yeah yeah, you sometimes you want it like in order to get like a really really really or a completely rock solid design.

A

You end up kind of taking the extremes of like how small could this possibly be to how much data could you possibly show, and then you sort of optimize for what's in the middle, but you want to account for those extreme cases where, where no, there was a time where a contributor actually thought about what it would take to remediate the issue and like wrote it down, and it was a full four paragraphs of like what to do um now. Do we want to optimize for that? Are they all gonna be like that? No, no!

A

Our contributors are much more lazy than that. That's not going to happen.

A

But yeah anyway, shubham good good. It will, uh when this lands man it'll be just the culmination of I don't know how many weeks of uh of effort to um make a highly a very referenceable. So one thing one piece of feedback: that's been given in the past by the way, if anyone's just sitting around not really taking notes, please feel free to drop notes into the meeting minutes. That would be really helpful as we capture these, especially for shubham.

A

um The one thing that isn't done here that you should give consideration to, and that is like the the potential combination uh combining of error, name and error code into a single column. Potentially, maybe, if those are.

G

Present they are in the single column, but what I've done is like when I merge them, so they were not looking very good like when they were just beside each other, so I just separated them them in the same cell. So one one data is, on the left, hand, side and the one and the other data is, on the right hand, side as yeah. So that's what happened. They both are in the same cell. Only.

A

Okay, so okay cool, so next time you present the this and you're and you're almost done uh um I'll.

A

Ask this so just as a I'll give a pop quiz, but I'll give you advanced warning of the pop quiz which is, can you just you know quickly, show an error that you might have received on the command line or or in the or from somewhere else, and based on the formatting of how that error code is shown copy it come over to this page and then do a find a browser find and if it matches a beautiful, then I have no room for comment.

A

If it doesn't match because they aren't together or hyphenated or whatever it is, then you might get the same feedback.

G

All right I'll, try I'll try that beforehand.

A

Yeah, okay, yeah yeah, and that doesn't mean that that that's the best and most appropriate way like there's some cleanliness and some readability, that's afforded by the separation of these two things. I you know, but you give but but give it some thought if you would or try try that out, because because you know that that's what people are going to do well, hopefully that's what they do. Is they like see an error like geez? What is this? They, they search, google um with that format match here, and so eventually you you.

A

Might you you'll find that if the two are combined, you would be able to get so fancy as to be able to say when they they're shown the error. You'd you'd have a system such that it. You can point people to docs.meshredio, slash reference, slash, error, codes, inc, you know, pound sign, anchor, the exact name and then they click it and they're brought like boom like exactly to.

D

um You know so think about that.

D

I

Cool anybody else have feedback.

A

Okay, what's next who's next.

A

And shubham, if you don't mind like just as a banner, oh thank you, that's great. What you might want to do is just try to recapture what was just said, and particularly sometimes you might just want to reject what was said as well. So like this is a good place to say to reject the feedback.

D

So cool josh, you're up.

K

So there were new ui changes, ux made for the sidebar now most were the hover feature and the change in the drop down and collapsible button, and one thing suggested, was to remove the background of the drop down. But I think if we remove the remove the background of the drop down, it won't be visible. When it's collapsed right, we won't be able to distinguish between the sub menu and the drop so.

I

Say it again yeah. um So uh if we remove the.

K

Background of the uh drop down the sub menu- oh yeah, it won't be visible in the collapsed state.

A

Yeah, it makes sense to me that you that we wouldn't or that there needs to be some other visual indicator, that yeah and a slight color change of the background of the sub menu items. That makes perfect sense to me.

A

It sounds like that was given as feedback to remove it previously.

A

Do you yeah, I don't. I don't know that.

K

Yeah, it was suggested to remove the background, but uh I told uh I replied that it won't be really visible. You can't distinguish between the sub menu.

A

Or others do you guys have.

J

I think we can leave some things to the users to write to uh ex. You know think uh so, since uh it's quite visible in the expanded state, so user may you know figure it out in collab state to some somehow.

J

uh But but you know, but if it is really very important to show uh show items like uh children's and parents inside uh inside the collapsed menu, then uh then what can be used because we can't use uh the offset right so.

A

Right right, even the office.

F

G

F

A

So sorry, even even the offset on the main when the menu is expanded, you, if you, when you click that I don't know it, wasn't there a minute a second ago that there was background highlighting.

D

A

That it's it's a little challenging to to tell that those are sub-menu items.

C

I think we are doing like more with the collapsed menu. Maybe we we should just stick to the original icons when it is collapsed, not showing what was like the sub menu options.

A

uh Yeah I, if we don't and then when someone goes to navigate to a sub menu option, would we then.

C

uh Whenever, like he hovers on it, my house on the collapse menu, it will get expanded. Second,.

A

Okay and then, and then you wouldn't show any difference, you just yeah.

C

And get back the original state where it was left like if menus were opened, menus are open.

A

Other thoughts, feedback.

J

Can you click on the patterns right now.

J

Okay, uh any item.

J

Yeah so since you are on the pattern speed, so I thought, like uh the configuration would get.

A

I don't have the link handy to the design and figma, but yeah. It's not clear. People need to see a difference between the sub menu items. It's not clear enough that there's a slight bump in people also need to see that difference when the menu is collapsed.

A

And yeah this looks great. The pull tab looks fanta, you nailed it, that's perfect. um We should get that merged. You should update your fork to your clone.

D

um It's the distraught thing. Oh, it's! The restructure. Okay,.

K

This is the sigma fight, yeah.

A

If those are the okay I'll I'll try to provide some final clarity, because it's just not that there's a couple of ways that it can be done right that one that's being shown the one just above, that's fine that that works. So people can't tell the difference, because it's colorized um it doesn't work when you close the when you minimize it, though, and so that just really wasn't very well considered for, and so it's really confusing. Nobody knows what those icons are.

J

Let's, let's take, let's stick with the current design that we have when we are in collapsed view uh with the proper. You know, uh change in the color that we are using in the currently and uh for for some uh in the expanded view, so we can go with the yeah this top design. So whenever whatever is selected, so this should be green, as this is right now.

F

J

F

Cool looks good or not, I don't know, I think I think so. I.

A

Think that I think that it would be if you provide that clarity abby on the collapse menu in the design itself. I then, I think, will be okay.

D

A

Can help as well but but if good emails have any feedback here before we go to navindy's topic.

A

A

All right, let me do europe.

B

Hi, so I just wanted to point out about the build and release meeting we have tomorrow. So many folks have been working on writing tests and uh making sure that we have automated tests and uh we have a test plan and we have. We are running tests and we are making measure ready for the v0.6 release, so I'm not quite sure about the status of the test, currently maybe push or someone else can point out how they are going.

B

But what what I propose we do in the next meeting is to get a status update on the test coverage and have the test plan in hand and look through look through whether we need whether we need to update it or whether there are changes. um Whether there are changes needed to the test plan and yeah and some other topics as well uh related to build and release pipeline for the get night out project, so yep uh maybe be used like. uh Could you?

B

Could you rally things around the testing you are doing and make sure that we have something to discuss by tomorrow?

B

Just dropped off okay, so I guess I will catch up uh async on slack and we'll make sure that uh everyone is aware, like the people who are initiating these tests or writing. Automating automatic tests know about the meeting and will join the meeting.

A

So raj brings up a good point about he's saying in his calendar. He doesn't see the to the thursday mesh rebuild and release meeting the one that's happening tomorrow, um so he brings up a good so raj. We should make sure that you have that this is actually a good note for everyone. um That's participating in the various meetings is that um make sure that you you're subscribed to the calendar when you, when folks join the community and you fill in a member form.

A

You're you'll be in you'll, be added to a couple of google groups. Those google groups are invited to some of the meetings, but not all of them. So if you're missing a meeting on your calendar, it's most likely that what you'll want to do is subscribe to the calendar and that way, actually, if you're subscribed, when new meetings are added or removed, you'll just get those updates without having to be explicitly invited so raj. If that doesn't work, let us you know, let let us know.

B

And lee like, uh could you mention like who else has been working on writing tests.

A

For so the so unit tests have been primarily within measure ctl the integration tests or um so um ashish has been. Who else has worked with you on adapter integration tests jared has, I think, taken a look.

J

Yeah, I can't remember the name, but after I uh so there was some I was about. I think it was uh aditya uh who basically there was uh uh a minor change in all of the e tweet s and b adapters, where uh uh I think I was working on uh super. He was working on the docks and then he needed to make the change. So he made the change in a little change in the workflow, uh so aditya has helped to make some changes and uh I have written the e2 tests.

J

I I can't, if I'm losing anyone's credit, please point that out, because I can't remember anyone else's name.

A

Make sense um and then mario had been kind of leading the charge on cyprus.

A

Nice all right, uh so we've got about nine minutes. Left, never do anything else so that that means tomorrow at about.

A

Two hours from now, I think, yeah, about two hours from now. In terms of the time of day,.

D

uh Aditya europe.

H

Yes, I will share my screen. Okay,.

H

All right, so what my issue was uh when I tried to create a pattern, I cannot see the kubernetes icon that is supposed to be here. I just see these two or whatever, only the adapters and measuring, but I don't see my kubernetes icon, uh even though I have I am connected.

A

H

A

Sorry, let me just let me let me stop you there. I think, um hey when's, the last time you pulled and built your ui is this from a build, or is this from uh the release.

H

uh On the slack channel, I displayed through the release the message system start, but this is one is from make run file. Oh okay,.

A

You might, if you haven't run, make run build ui in a while. um There was a bug that was fixed if and it pertains to what you're just showing if you go back to the patterns, if you go back to the configurator.

A

When you click the selector for the system that you want to choose components for uh this one yep, if you would and then yeah when and in the upper left-hand corner, when you choose the system for which you there's a bug in which the z-index is not being uh paid attention to very well like and maybe- and maybe you don't have kubernetes there at all. But I wanted to point out that it might be okay yeah. So you don't, but I think abby that bug was fixed, correct. uh The z index book.

H

Yeah, it is next okay, I fixed that, but I don't know why I I I pulled that lettuce after that fix, but I don't know why it's not true.

J

Yeah, you have to build ui or either run run ui dev to see the latest ninjas by the way, but.

H

Still the kubernetes icon thing, maybe it's maybe different from the.

J

Bug is different yeah, so so I think uh ashley is actually a good uh guy to answer yeah, uh but, and that is the same problem. I am also facing uh that our problems are registered. Registering cube api open, schema right now, so.

J

We both are, on the same. What that's what I want to say.

H

D

H

I tried the link that ashish mentioned in slack at this one, but yeah I'm I'm shown with an empty empty added.

J

Okay, can you try uh service dot key? It's also. I mean this one. I don't think it would make any difference, but uh instead of just paul just service capital.

H

J

J

E

J

So that the config is reaching initialized yeah, can you click on that chip? Can you show me the logs now messages our logs.

H

uh All right so messy sure this one yeah.

J

Yeah, so can you do control shift tips and find basically just type looking? I think that's one of the locks that comes along with that. So can you expand it? Actually, it's too small to see.

H

Okay, uh I'll expand it yeah.

H

Yeah, what was I supposed.

J

To say, uh capital, l and then basically looking can you type looking and see if it is press enter.

H

Yeah, it didn't find it.

J

Okay, um okay, so this shouldn't have happened.

J

Have you done okay, so you have run, make run fast? Okay, yes,.

H

Yes, yes, make run fast.

J

Okay, actually I mostly don't make run 16 fast and I'll never face this issue, so maybe in the in the uh make file. Maybe we have because there is an option to escape the kx component generation. So I'm just looking at that right now that whether we have skipped it by mistake.

H

Okay uh but yeah, I I don't see the gate icon and from the release that message system start off. I don't see that icon.

J

Okay, so uh I found it uh basically uh in the make run fast, uh but I think it was by mistake with something and some due to some merge conflict. There is a skip component generation which, which should only be present in maker and fast, skip component generation. It is present there, so it is just a one line fix I'll have to remove that one line. So after I remove that line uh actually what's happening.

J

Is that because it takes too much ram to do that, so we have an option to completely skip it and by mistake it made its way into this uh particular recipe, so I'll have to remove that.

J

Okay, that was that was a one-liner fix. I was like I was like hung on it on for, like four five hours.

H

A

Great okay, we've got so we got a an issue to file there, a bug to fix anybody else. Have there were two other topics for today? um Does anyone else have anything else before I touch on these for 30 seconds.

A

Okay, there's um well, messry does a lot of things. We don't have many videos showing people what those things are, and so there's been a number of people and at work trying to capture some animated gifs or some short video clips to help share.

A

You know, share the features with with everyone else, so by the way, if that interests, anyone on the call please do it. You know please engage like let's get some, let's get some better examples um shown to people what the features do.

A

One of those efforts has been um to create a small.

A

This is just a 15 second video and is really needs to be better than this, but but it's just a short clip that shows like one feature and so um I'll. Let you all play it um for yourselves. The link is in the meeting minutes, um but it was just drafted um late last night and so give it a give it a play. You know certainly give feedback, um it can be, I'm sure, improved in any number of ways.

A

um It's just kind of showing how we might tell people about one feature, and we would go on from there to tell people about more um something. Okay, so so yeah be very critical. You know any any feedback. All feedback is good that was made in adobe after effects, so the source files are can be made available as well. For anyone that wants to iterate on it and try, you know, try their hand at it.

A

The other topic, as we go to wrap here was, is the service mesh playground. um This has been just an early topic coming up in on the monday.

A

Meetings in in the websites call is the notion that well we're it's about time to uh design the the playground, so I think I think that the thought right now is that this would be probably be played up mastery.I and there would be a hosted instance of measuring for people to come, learn about service meshes and do so through the auspices of measuring- um and this is a this- is a an incomplete draft. There's it's yet to be a first draft of like what that is.

A

It does give a little bit of um information about some of the things that we need to do or how we would approach that um not much so so again, just a quick, uh just a quick word that hey there's an upcoming project to do it's briefly briefly described in this spec.

A

There have been a one or two um designs given as to what its main page might look like.

A

And so part of the concept here with the service mesh playground is that um many of you have participated in creating some learning paths for people to come and learn about service meshes and that in those learning paths, there are interactive labs that people can take to learn how a service mesh works. Those la those interactive labs are held are hosted on catacota and they are confining as to what all the user can do, and so the idea with the playground is to give them much more much deeper.

A

Access to the infrastructure and larger labs ultimately allow people to certify as being service mesh knowledgeable.

A

That's what the initiative is about the in the well we're going to have to go. Look at the history of this file. It looks.

H

Yeah yeah, I also looked it yesterday and I could not find the design because we just find the logo.

A

Yeah, there's no pull request system here. It's just anyone who has edited access can go so we'll have to go, find those but um but yeah. So it's an upcoming project. So does anybody have anything else to add or comment before we go to close out today.

A

Okay, I'm just hunting for those mock-ups, but if not uh so tomorrow's build and release meeting, which um I think we need to go update the calendar it looks like it looks like that needs to be moved to an hour earlier on the calendar, but um ashish piyush looks like you all have topics, mario has a topic uh for that that meeting tomorrow and yeah. This is also a meeting. The building release meetings, also a good meeting for some of the discussions around github actions.

A

So um actually we might want to ask hirsch about some of his what he's done recently.

A

Anything else today.

A

Nice, okay, it's nice to see you all um happy hump day. I guess wednesday.

A

Okay, see you all tomorrow,.

H