Layer5 Meshery Development, 1 Feb 2023

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Meshery Dev Meeting (Feb. 1st, 2023)

Description

Topic:

[Pranav] - [User Extension] Credentials
[Ashish] Fix permission issue in docker container
[Ashish] - Failing E2E tests/Call for build and release call.
[Ansh] - “Material UI Table Pagination [#7009]”
[Abhi] Fix SVG scaling Problem in Folder: MesheryMeshInterface [#6966][#7035]
[Abhi] - [Confirmation Modal] Add triangular badge in Confirmation Modal #6986
[Karan] - “Fixed Svg styles in MesheryMeshInterface [#7034]
[Subedy] - Meshery Catalog updates
[Gaurav] - Meshery UI and darkmode issues to take up - Issues
[Uzair] - Update meshmodel to add support for defining enabled/disabled relationships. #7005
[Gopi Vaibhav] - GitHub Action to tweet on label - PR Link
[Rudraksh Karpe] - “Fix compatibility matrix order issue PR-LInk”
[Abhi] - Playground and MeshMap beta signup
[Lee/Gaurav] - LFX Internships
[Future]

[Lee] - CNCF TOC supports moving Meshery Playground to CNCF-hosted infrastructure

A

A

So welcome everyone.

B

A

Everyone to mystery development meeting today is.

B

A

Of Feb with that um I, don't think we have newcomers on today's call, so we can directly go ahead with today's topic, so we have a good list of agenda items lined up today and I'll just share the link of meeting minutes again in chat. If you haven't written your name in attendees list make sure you write one so with that first year we have Ashish on couple of items, so Ashish would like to go over.

C

D

Okay, uh I guess. um The first item was taken care of we already discussed it. I, don't know if you want to discuss it again. um This was about the UI changes, but I'll come I'll come back to it, so the first thing was um so I I'd like to give a little bit of context on what I'm talking about. So we have this machine on message.io catalog. We have this uh catalog of all the items.

D

So here you have designs on also you have wasn't filters and there is a workflow which actually adds these things over here and that workflow is over here inside of messy.io repository instead of here there is this workflow update, catalog and the it's of type repository dispatch, which basically means that there can be a post request sent to a certain endpoint. So this is basically uh for any trigger that exists outside of GitHub uh for any external web hook. We can have this particular workflow run from outside.

D

In our case, that outside is mystery.ler5.io, which is Mystery Cloud. So every time a user publishes a design.

D

um This particular and I'll also run mystery server alongside too so that I can also show what it means from the UI side, what it means to click on, publish and all so, when someone clicks on publish this particular workflow kicks off and uh the what this particular workflow does is. It goes over to mastery.layer5.io pattern, slash catalog, and over here you have a bunch of, as you can see, I have these catalog items. These are designs and similarly you can go to filters, slash catalog, so e.

D

For each of these catalog items there is an endpoint uh for patents as well as for for filters, and it goes to this particular endpoint. It takes all of these uh designs in this case and it publishes them over to the creates markdown and all so that they can be published over here on the website. Okay, so that was that was the published workflow. There are two parts for this published workflow. One part is how So. Currently there is no.

D

We are on the way to have a whole workflow where people anyone can publish anyone can act at least request to publish their designs on memory. Cloud right now, only um people with privileges of admin and privilege, people with privileges of curator, can do that, but um later on, users would be able to request for it and there will be a queue from where the curators would be able to approve it. So that's the left part of the flow, which is to say uh when and how someone can publish the design.

D

What I'm talking about in this task is, after that design is published. What can we do with that? uh So just to give you an example: I'll open the UI, so I can show the way from where any user in future any user, but right now only admins and curator can publish a design. So if you log in you have to be logged in as remote provider.

D

So if you log in as remote provider and if you go into designs, you'll find there is this button, you can use this button to show you the catalog items or hide the catalog items. So, for example, in this case this is the catalog item and you can for all the private designs that you have.

D

These are my private designs, so you can click on publish and you can fill this form and right now, because I have the Privileges of an admin I this these will be directly published, but later on, the user flow kick off when someone requests a publish. So after someone publishes a design, what what happens is that this particular endpoint is hit and all of these designs or filters or whatever catalog item is they are fetched and they are added. So this is an additive change. It does not.

D

uh So this is kind of this actually Edge triggered, which is to say this workflow that I just showed you. It is triggered only when an update takes place onto an already published design or a new published. A new published design is added. So in these two cases this particular the workflow kicks in, and it makes sure that the published design that are available on this particular endpoint are also available on on here.

D

It doesn't do it one by one, which is to say, if you add a published, so the the first item that that wasn't under discussion was comparing two approaches. The current approach and the latter approach uh there are pros and cons to both the current approach basically goes to all the um fetches, all the designs, all the published designs and then publishes them to the catalog, and there are no duplications. If the IDS don't change.

D

The drawback is that if we have like hundreds and 200 published designs in that case, this is not really a good approach, because if someone changes, if someone adds one more published design, it would still go and fetch for all the 200 designs and do that change. But since currently we don't have that many uh catalog items, uh it's fine. If we go with that approach, if we go with that current approach, uh it requires less rework.

D

The other approach would be whenever someone adds a particular design, then that ID is passed on to this particular workflow, so that when this workflow kicks in it actually fetches up the design of that ID and then basically on the catalog, only that design is added. There are cons to discuss. There are multiple cases where something can be missed, so I won't go over them. I'll uh I'll come to the unpublished thing So, currently with the unpublished approaches whenever a particular design is unpublished or deleted.

D

In both of the cases, we would want to delete that particular design from the machine.io catalog. So there is a PR up for that, and uh currently uh there was some the logic wasn't pretty great previously, which is on delete and on delete as well as there was no flow for unpublished, and even if someone deleted it, the same workflow was run and this workflow actually does not remove any existing catalog items due to which we had a lot of duplicates over here.

D

So what this purple request adds is it uh it has it basically splits these things into two functions? What one is the dispatch catalog sync synchronization or dispatch catalog ad, so whenever someone adds a particular design, then the original workflow kicks in. But uh if someone deletes or unpublishes it, then the them. Basically, if someone deletes or unpublishes any pattern or design, then the workflow is kicked with a different, so the body is different for that request, so the web hook. So this particular event type is different.

D

So there is a sync catalog event: type for or add catalog. This can be changed whenever an ad event or a publish event is triggered, and then there is a delete, catalog event type when a particular design or particular item is deleted and in the client payload we'll pass the particular pattern ID or filter ID, and since the file names inside of all these files, name inside of the message.io repository are such that they have the the uuid as the file name.

D

We can easily find out which particular file to delete so unpublishing will trigger a delete over here and these files, the particular items, would be deleted. So he said uh that was. uh That was my first update about publishing and unpublishing. If someone has any questions, I can answer them and uh then I can move on to the next sub point of this, of this particular pull request, or this particular change.

E

I have a question for you. um So as far as the numbering of the um items in the catalog, where is that, is that um coming from the workflow, um because there's duplications in the numbers and I was I haven't had time to look over it. So I was curious if you knew right off the top of your head, uh where the numbering was coming from, because as far as I know, it's not dynamic.

D

Okay, so if you're talking about this message000 yeah okay- so these are not coming from the workflow, they do have a unique uid, but those are not used on the UI for some reason and I think we can get rid of this mesh. These are completely uh client-side things these are not coming from, but we can still distinguish them individually because the file names had to have uuid.

D

So if you go into this measure.io repository in inside of uh here, you'll have catalog, and if you'll go here, you can see that the yaml file, uh the file names, are actually the uuids, but these uuids are not shown over here, the basically the markdown that is generated in the workflow. It I mean I'm, not sure about this. Maybe someone who has worked on the UI can tell me about it, but uh what I saw was that these meshd 0 0 and mystery 002 and zero zero one. They are randomly added, so they yeah.

D

These are. These do not necessarily help here to index a particular design or a particular catalog item, so we might remove them or um that there might have been some logic. Basically, maybe there would have been a logic that whenever a design is added, uh these counts are incremented, but there is no consistency from the UI side, so yeah I I yeah. That too, to answer your question.

D

B

Yeah, so we we actually just talked about this this last hour, yeah.

C

B

Those uuids that are so the unique identifier for each catalog entry needs to be some form of it needs to be referenced here um and the the issue is that the uuid is terribly long and doesn't isn't very palatable by people perusing, the catalog uh and so Colby. Thank you for bringing this up, because this is. This is an item that you might be able to one figure out a solution to and then two implement the data is there like.

B

Ashish is showing, but there's there's the the probably the the challenge that stands in the way or that maybe the toughest challenge is just the ux of the thing like it's kind of nice, for people to or the original thought is that it's nice for people is that the original thought is that the catalog is supposed to be contain a number of best practices um and that those best practices might be numbered, or there might be a you know, a unique reference to them that they might come in um different forms, so one of them being a pattern, one of them being like a filter, maybe configured in a certain way or just another.

B

One being and like an Opa policy or an ebpf program or and yeah that they would each carry a small little moniker like for the patterns. It would be like a the pattern logo of meshery and then some some unique identifier. Some number now we can Auto increment, like starting from zero and kind of counting up, is nice for us humans to be able to say: oh yeah, it's you know number 37.

B

um how that having said that, it's a lot easier, programmatically for this for meshery server to generate a uuid and track that, as part of you know the unique identifier for the design.

B

So then the conundrum is, you know, is like how to make that concise. How to give like. So if we were to just take the stuff, the last suffix or like curtail the last five characters or so off of the uuid a would, we generally expect that to be unique enough, that we could have like measuri and then five characters and B is five characters enough. Is that if five characters too long for us humans to really make use of it anyway?

B

So uh these are the questions sort of surrounding this yeah.

E

I was digging into it um and trying to find out where it was coming from, and so that's the reason I asked kashish was: is it coming from the workflow or is it coming from? The logic of you know, measuring communicating with the the UI the front end and Antoinette might have some feedback on that as well, because a lot of that is like front-end stuff. So.

D

Actually, if you'll see on line number 71 here, this particular pattern, ID is what you you see over here, so actually zero zero, four zero zero! Whatever- and here you can see, this is generated on the basis of this idx variable, which is basically an iterator inside of the for Loop, so if you're I, so the six or so some, these six particular patterns that you see over here.

D

Why, when this goes over each one of them, it just adds adds them as much three zero zero one: zero zero: two: zero: zero: three uh zero zero! Four like that.

E

The tricky thing is that there's the wasm filters, then you know there's like several categories of um patterns basically and um yeah, so yeah I'll I'll look into it some more Lee.

B

Oh nice yeah, oh.

F

Yeah I, don't know exactly so I I wanted to add on to that um the when you were talking about the Metro, but the ID numbers. So on each of these markdown files. There is a pattern ID that has that and so they're all multiple with the um the duplicate numbers. So it's in the collection underscore catalog deployment and each of the markdown files have the uh the pattern ID or filter ID or whatever it is, but it's defined there so um because it's already defined, though shouldn't that have already been trained through the workflow.

D

Yeah that that is defined, uh that markdown was created by the workflow and uh that ID is given by the workflow over here, and this ID is currently just a random. As I said, it's like one, two, three four, depending upon where the iterator is, while going through all those particular patterns. But as Lee said, this can be something. uh This should be something more unique like you, a part of the uuid or something like that.

D

So the change needs to be made over in this workflow on line 71 and the workflow actually decides what this pattern ID will be so uh yeah.

F

Okay, um okay! Well, then, I just wanted to point that out, because there's still uh when, even when you look at the catalog, we still have these same different number and it's still the same number in the market. uh Mark I don't follow so um okay, yeah I agree with Kobe. We still need to look in into it and same thing for us to uh trying to make this better. So.

D

Oh okay, thank you for the feedback. I'll uh I'll collect responses based on that the ID thing and make appropriate changes after I make the unpublished changes. The second thing I want to I wanted to add, was currently we have filters and Designs over here in the catalog, but uh one of the other things that you also have are applications and applications like kubernetes or Helm chart or your Docker compose applications, and these are the easiest to import applications.

D

So if you don't want to design by scratch, you can just import them, but we have no way to publish them right now and I. Think if you have, if you can publish design um on like in our internal system, the format that we use for applications is also design and I. Think it's more like a proposal that we should also add a publish, because it's it's quite straightforward. So, just like we have publishing of designs, we should also have publishing of applications.

D

Now it's completely on the user facing side how how we would want to show it so would should that be like indistinguishable from designs, because internally they are saved as designs.

D

So the same exact workflow can be used, and over here in the catalog, you wouldn't be able to distinguish whether it came from a application or it came from a design or a better use or a better uh ux can be that we actually show it somewhere that actually, this came from an application and when user clicks on that, they can download the original Helm chart or the original kubernetes manifest from which it from which it came from so yeah. This is.

D

um This is more like a proposal, an open consideration or as to how we would want to if, if we want, if we went ahead uh to publish applications, how would we want this particular catalog to look like for applications and what should be the user experience around? That.

C

D

S they can make.

B

Don't don't do it, okay, we're not trying we're not trying to recreate artifact Hub or another Helm chart repo, like let's add novel value. It would be my suggestion. I did how you how you arrive at that. That's uh you're you're, correct in what you'd said, um but if you think about you know, if you think about other well, artifact Hub or some others escape me, there's a a couple of other chart.

B

Catalogs um I, don't know it's one of those things to like put a pin in and come back to after a v one zero. If the catalog gets enough traction and people are using it a lot and they find it just that it's um some of its features as a catalog are quite helpful and they find that they're using it for designs and things. It might be that requests come through to be able to also publish um Helm charts there.

B

It's just that at this phase of where the catalog is and measures use is that kind of it wouldn't be a novel capability and we would have to compete with other projects that already do something similar. So it's anyway, that's the logic that I would use.

A

Cool so yeah I've copied the session in meeting minutes. So let's move to next item, which is by harshit on Final review of dark mode before merging it today so harshit, are you ready with your update.

G

uh Hello, yeah, I'm, audible, yep, okay, so yeah I'm sharing my screen.

G

So yeah I mean yeah, here's a dark mode and uh and I think final query was to resolve or the merge conflict and I did that yeah, so yeah I think we are done with all the previous issues and yeah. We are ready to go now.

A

And conflicts resolved on the extensions as well right.

G

uh Yeah, it is.

A

Yeah so I'll be or anyone do you have any final comments before we hit merge this.

H

Yeah uh I think uh we had a good number of reviews for the dark mode and uh if, if there is some uh remaining items uh left, then we will better, it would be uh covered in a new beer so that there is some amount of progress uh being done here. So I would I would go to uh I mean I'm in favor of merging it uh right now because yeah we have reviewed it much many times and yeah.

H

There is no such big red flag to haul this beer on and if we halt it like we'll, be just bumping, the number of merge conflicts so I don't think it is yeah worth hurting.

A

Yeah cool, so let's merge them and we can follow up on the bugs or issues.

G

Yeah one more thing: I want to add here that if we ever I mean if we move to the mui5, it is also this technique will be also compatible with that also I mean uh yeah, so we can I mean we can I mean we are comfortable. Also, there.

A

Yeah, that's great so yeah we have so much so that's cool. So thank you for your update. uh So let's move to next item, which is by Lee on allow and allow or deny selectors in mesh model component.

B

Yeah is there on I know he. He was pretty hard at work earlier, but I had to go get dinner, so yeah I, don't think he is um I asked because it was there. It was part of this, a conversation earlier today on just an augmentation to mesh model, um and so there is section here if you were to one, if you were to just adjust your zoom level, maybe not so much for me, but I spec for others.

B

um There's a section here that has a schema for com relationships, I, think it's the relationship schema. So it's just there you go. There was a discussion earlier about all right, so just extremely briefly: meshery as a infrastructure manager models, the elements under management using um mesh model and different elements under management inside of mesh model are represented at objects as resources, and each of these objects can be well. Some of them can have relationships with one another um and some of them can't or shouldn't.

B

So, if you think about an object on management being something like a kubernetes service.

B

A kubernetes service might have a relationship with a kubernetes deployment and that relationship might be network based sort of a network-based relationship. So.

B

Great, we need a way. A mesh model provides a way of expressing that the fact that there there isn't a relationship that is allowed um between kubernetes service and kubernetes deployment. Now not all elements under management should should be allowed to have a relationship or it or does it make sense for them to, and so there are different types of relationships. The example that I just gave is a network-based relationship.

B

So if we continue on that same example- and we were to say so, should a kubernetes service be allowed to have a network-based relationship with um with a kubernetes secret? Probably not like Secrets. Don't really, you know, accept Network requests not like a deployment might, and so you know that's just inappropriate.

B

um So today's discussion was to augment this model with the ability to to be a little more expressive in the current version of these schemas, which, by the way, these schemas and their current version, there's an open pull request that init them that that Define their initial schema. So please go review that open PR. Actually, there's there's help needed on uh on refining the schema. So this is part of that is refining it, so agarav, if you were to scroll up just slightly.

B

um This particular schema, is for the mesh model construct of a relationship, and so this says here's how you can describe a relationship between one or more elements under management and we we're using kubernetes um constructs as an example and we're using in this particular case. It's a relationship of type Network and so under the selectors stands under. That section are selectors to allow a relationship and garv. If you scroll down a little bit, there's now a section for allowing so you to identify um anti-selectors or deny selectors. So in this example, I was using.

B

It would be something like explicitly saying: a kubernetes service cannot have a network-based relationship with a kubernetes secret that expressiveness the ability to Blacklist a relationship deny a relationship um wasn't present before, but that was just a discussion happening earlier today. I wanted to surface that, for you all to consider the terminology that we had considered using was whitelist and Blacklist, and- and we were sort of landed on, allow and deny, as hopefully, equally meaningful uh denotations of when you're.

B

When a relationship is allowed or when it's like explicitly disallowed, there are other there's a number of scenarios in which you can imagine. It's very helpful to be able to express the fact that one component wanna can have a relationship with another one or it shall not. It can't that just wasn't part of the spec the schema before.

B

So yeah, so this needs to be added to the openpr, assuming that folks agree that that makes sense that you could be more expressive. Yeah yeah.

B

That's the topic of you, know, follow-up and slack and or a follow-up on the pr with with questions. Part of the help that's needed on the pr is really something that anybody here could do the and you know what I'll I'll point it out separately. So we can continue I'll point out and slack some areas that people could help on.

A

Yeah cool, uh thankfully, so if anyone wants to know more about mesh model and how we are defining relationship schemas, so you can give a brief read to this talk. I have shared the link in chat so with that, let's move to next atom, which is by internet on decision to make mui 5 V5, upgrade in dot 7 release.

F

So, um for my part, we decided to operate to remove uh version 5 in Euro yourself, um the UE uh directory and also mouth yourself, um since they um completely well actually, since map completely depend on Metro yourself. So any changes that is made- and you will also have to be a change over the mismatch.

F

um So we decided not to move forward at this point because we know it's a lot more work that needs to be done and requires more uh Hands-On and more effort to simulate this happen. However, I still have an open issue. If anybody would like to add on to it um my suggestion um I would be changing it.

F

Updating it all the time on this particular open issue to Colorado, see if anybody would like to join in on the effort, um but we will be starting this, probably um maybe when, when at the Pacific date, you want to do I know. You said you wanted to do it uh in this year. But when would you like to.

B

Yeah, um the bunch of um user experience user ux designs that, if we're going to put in the effort to rewrite UI, then we would at the same time want to include a bunch of ux changes, because there's a bunch that have been teed up that we haven't gotten to yet, and so the most appropriate place to start is with ux designs, going through those reviews and then getting to um uplift.

B

You know upgrade and re-implementation so like not this year, like like literally not within six months, probably like um it's just not so it's something, that's well beyond the scope of what was planned for v070, unless you all have friends who are talented and have time that could change, but of the folks that are engaged right now. It's just I'm not in the cards.

F

uh Yes, so uh to add on to that, the on another thing that we were able to accomplish, we were able to migrate, providing users to um at least react, 18, knit 13 and movie version 5..

F

um So from that changes that was made, that can give you an idea of what to look for in yui and also a mess map. So that's where we are on on the uh the funding part I, don't think anybody have any questions or would like to add on to it.

E

Yeah brief question: um so if we were to want to work on this particular PR, we would be working on the branch that you've got here in GitHub. Is that is that right.

F

uh Working on which one.

E

So you have a PR, that's open um here on you know six six eighty um I mean if we were. If we were wanting to contribute to that particular Branch, then I'm, assuming that we would be working on that branch.

F

Oh so I have an open issue, but not a PR. We did have a PR but I closed it, because once we started um when Aviano started working on it and then it was, it was at that point. We had to stop so the um but the issue here I still have it open and um if you would like to add on to it, let's change it to the um the note that I made, then yes, you can, but there is a.

F

There is a another uh branch that has started the effort, but it's too far behind where we can even merge that to the current um uh branch that we want. However, as I mentioned before, that is something that you can use for. Information is, is redesigned into where you can see different um uh layout. um You know you have your typical dashboard, headers, so on and so forth, but it also using the the newest um Redux toolkit and also um the in-between uh movie version 4 and version five, but from where they are.

F

You can easily change some of it to be completely. Yes, um the link that Lee just put that's the one that you can take a look at. So it's already migrated to use video toolkit. It's already like close to version five. It just needs to you know, remove some of the um the um the Legacy package, which is Moon Styles, they've already started.

F

This particular package is not going to be supported in react 18, so once we make that jump to 13 and 18, we will have to um leave that behind, but um and even then this and move it won't. Let you use it anyway. So, um but yes, that's something that you can look at um see if there's anything that you feel like that was already done.

F

That could be um added to this current, but but we'll have to see, um but there's a lot like what he said uh create the mock-up um to go over the designs effort. um We also need to start working on converting some of the react.

F

Flash components to function component um I, even I, would suggest that right now, if you were to make any new um new uh component, make sure it's function and that class component, unless you have to use Flash components and that's fine, just let us know and then we'll we'll go from there, but um it would just make migration easier uh later when we get there. So uh did that answers your question Colby.

F

I think so: okay, all right yeah.

A

And yeah with respect to that, I would like to add something. So we do have our docs of uh Redux. We are using read textile State, Management tool, so I'll just share a link of that in chat, and we also used to track the implemented UI items over here. So this can also be referenced.

A

Yeah thanks Antoinette, so I don't think we have more questions. So, let's move to the next item, which is by Avi.

H

H

Specific issue where we are talking about fixing SVG scaling for uh problems, and since it is uh like a quite a big effort, because there are a lot of folders and files in the Machinery UI itself, so I would like to take an approach where uh we would be taking one of the most important folders, where uh these changes would need to be implemented first and foremost and I would start with the measuring mesh interface. So what measuring mesh interface uh basically is having is that it is? It is managing the a design configurator.

H

So this is the design configurator. It is managing all these uh Pathways. All these areas, and especially uh rgc forms so since rsf forms are used at two places. One is in this machine UI and another is in the mesh map, which is an extension to machine so uh yeah. We need to start with that thing, uh so the the ask is like quite simple and uh I would label it with a good first issue or uh yeah, so something of that shot.

H

So uh what what we require is that, and the same I mean the exact effort has been made uh earlier also in, and these two commits describe that effort and uh what it is required is for each SVG be used uh where we are importing it from the another Library where it can be material by library or font or some or maybe something else, but we usually use muized icons and font.

H

Awesome icons, uh yeah, so in in that, in these two cases like whenever we are using any SPD from any external Library, so we would be, we would have to be specifying a height and width prop uh just to get rid of all the scaling problem that we are facing right now right. So, as I said, like these 15 files have already uh been covered with the I mean the required changes and uh I would like to ask for volunteers, to you know: do the same changes like uh kind of uh yeah.

H

It will be a good opportunity for you to actually uh dive deep into one of the most important code base. uh We we think uh inside the measure UI. So here is the yeah, so so I just opened the commit I think so yeah so a bit. No sorry also this. This is the. This is the file where most common icon, height and width properties are listed down. So, for example, small icons are a given height 20 by 20, and so as medium icon are given 24 by 24 So, based on.

H

What's the required of the requirement of the UI and comparing with with I mean, what's basically based on the requirement of the UI you'll, have to choose between the four uh sizes and uh basically use it so using using this uh Central interface will allow us to consistently provide uh height and width prop to all this, which is that we use and uh and and thereby will be getting rid of most of the problems uh related to SVG scaling, so that was a minor ask and yeah I think that's it.

H

So if someone is interested uh in, you know doing it, so they can follow up on slack or in the chat or just talk to it. Right now,.

A

Yeah, thank you so moving on to next item, which is like for me so I also have a similar topic, So, based on your request to update the to use the new SVG system in mesh model. So I, like merged up here, I created a PRN motor using the new SVG system in rgsf. So yeah I followed the same approach and used the icon medium, but that was looking a bit bigger, so I switched back to Icon small on rjsf.

A

So if we take a look, quick look at the file change, so I have only edited the custom fields for rjsf that are in mystery, mesh interface, so some of the files are done and rest of the files can be updated. As in your new issue,.

A

E

So any comment will.

H

A

H

No I think uh yeah, it's pretty straightforward. Thanks for doing it.

A

So yeah next up, I just want to ask about the failing, Cyprus integration test. So I it's with the it's with respect to the next upgrade I. Think so are we tracking this request anyway?.

H

I think I think we were talking about it in one of the slack conversation in measure UI channel, so I don't know where it went uh at that time, but yeah we somehow lost the lost the context so yeah. We need to create a specific issue for it and that needs to be worked on promptly.

A

Yeah, okay, I'll create one issue, and then we can do a follow-up on that.

H

Thank you yeah, so.

A

Let's move to next item, which is browser, but uh I, don't think it was. There is on the call with us today.

A

So we have ayush with two topics.

B

So you know uh sorry interrupt I'll cover uzair's topic for him since he's not here, and that's just um to dovetail a little bit off of what we've been discussing around uh publishing content into the catalog which, by the way, there is a visual workflow inside of meshery's architecture, deck that outlines how it is that content is published. The content publishing flow goes was originally to be done. Well, it was originally done by people, editing, meshery.io and adding their designs there. It's um that's okay, it's not quite convenient. It makes for everyone.

B

It makes people go to do a PR and have reviews involved and um we desire for a more dynamic system in which people can not only consume from the catalog programmatically within their Mastery server, but also that they could publish programmatically and to facilitate that publishing. We really need to know you really. We really need to know who's publishing and have a link back to the individual or the entity that has published content.

B

Eventually, content will be classified as like community supported, supported by a particular individual or a particular organization, if you think about Docker hub or other hubs there's. This is just pretty commonplace stuff, you have a if you have a catalog full of items.

B

um You know the there needs to be a bunch of metadata around those items who published it where it came from. Are they supporting it? What version is it at um what technology does it apply to what it? What versions of that those Technologies is compatible with, and so on? We have a lot of that information covered in Mastery server today, not all that is yet exposed into the catalog, and it needs to be measuring the remote provider which we've referred to as Mastery Cloud um yeah, just to be involved to provide identity.

B

So so, if a requirement of the catalog is to ensure that we understand where this content is from that people can get credit for their content, that people can support their content or be contacted about it or or refer to theirs.

B

um Well, that has to be need to have identity, Associated and identity, um as it is built into mesh as it is not. The built-in domestry server today needs to come from a remote provider. That process of curating needs to be involved as well. So people don't publish, swear words or other nonsense that doesn't make sense into the catalog, um and so that's a little bit of background for some of the questions that you all are asking in slack in other places.

B

um A lot of this is written down by the way in the extensive list of design specs that are in the um Community Drive uzair's particular topic had to deal with um use of three different um States for a particular design. If it was private and owned by and created by a particular individual, they have the choice of moving that from private to a public piece of content, as they do that it'll go through this.

B

These workflows that we've been talking about and so um trying to build towards some of the things that Ashish was mentioning earlier. Uzair was just making some changes around those States private public, published Okay, so just kind of a he's been participating in helping move. That forward.

B

There is I feel an LFX internship, probably coming on for the catalog itself. It has been the subject of an internship, previously I meant to put LFX internships as a has a topic for today, as well as gsoc, because we're going to go ahead and go we're going to go ahead and open up a number of them um well as soon as we can, and one of them will probably be on this topic.

G

B

Anyway, that's about it.

A

Yeah cool- that's great! Thank you, Lee for giving an update on public pattern on public catalog. So next up we have uh ayush on uh improving Michigan and logs for registering workloads. So ayush would like to share your screen.

A

Do we have a use on the call yup we do foreign.

B

Well, maybe after if Alberto has an update today, we should probably let him go since he also has open PR's on adapters. But if not, if that puts Alberto on the spot, um then maybe a she sure internet, you could speak on behalf of ayush who's, maybe not with an earshot of his speaker.

B

Alberto. Do you want to give an update on your progress with the Mastery adapters.

C

uh Sure yeah I have.

B

C

Prs, open, I think uh I think there's like four or five open one for app match map mesh other four um nginx, one and I. Don't remember the other three, um but yeah I'll I'm, still waiting for uh like more comments uh regarding those. If the, if I need to do any changes and uh regarding the other thing, the arm.

C

uh The arm issues that have been uh like investigating uh uh I mean just a heads up and that I'm going to put that on pause for a little bit, I'm going to be like traveling, so uh I won't be able to make any updates there. So yeah just says.

B

Heads up yeah yeah nice, but but will you be out for about a week or yeah.

C

B

A week, okay, yeah nice yeah by the way of the latest progress that you've made on the on arm compatibility. Remind me again. The Crux of the challenge that was coming down to was part of that in involvement of wrk2.

B

C

Lot different: no, it's mainly due to the uh julipsy container that we are like putting everything uh at the ends.

C

um Yeah I'm, like searching for a way of uh um like including I'm, compiling dlfc libraries on on arm I've, been making some uh like uh tests on our machine that I have uh but yeah I'll I'm still like looking into it, uh because it's uh I have to build a APK package for uh for Alpine and I have to do a bunch of things in order to make make it work. The available options are not there. I mean as far as I can I could like find.

C

um There is no like out of the box solution for uh glibc on our jet. uh Well, some people are asking for everybody. That's about it.

B

As a refresher, um Ashish Lipsy was, is the requirement for or is necessary for, measuring servers use? Sql Lite is.

F

That right, yeah.

B

um Now might be an appropriate time to remind some of us about the efforts that you put into measuring servers, support of an alternative database other than sqlite.

D

Yeah, so there is an open PR in missionary, which was made few months ago, but that was put on hold due to some other reasons and that pull request was to have a postgres as an alternative. So we would be later migrating completely over to the postgres because, as we have things like workflow engine coming on, uh which which the like one of the workflow engine is temporal and temporal uses postgres. So we would so using an external database actually solves a bunch of these issues.

D

So since all the database stuff would be handled by posters container, we don't need to have all these internal dependencies with some external shared objects, like we have in case of sqlite right now, so uh yeah that that, uh and because of that reason the other issue of arm was put on hold four months ago, because we thought at that point that the postgres pr would get would be land would land and once it lands, it would make the previous PR uh put the issue obsolete.

D

um There would be arm compatibility but yeah, but currently we are at the same page. We were like at that time as well to find the appropriate dependencies and, as Alberto pointed out that yeah this kind of struggle to find the accurate image and I he has. He has created his own image as well, but there are also some dependencies are missing so to find the exact dependencies. So even in glipse there are a bunch of other dependencies as well uh like even under that umbrella, so to find the exact dependencies that would not.

D

That would not cause that that would that will allow us to run the Mastery binary for arm without any issues. It's kind of a challenge right now, so.

I

A

Am I Audible yeah? You are ayush.

I

Oh sorry, I I, don't know why I was not able to share my screen.

A

Yeah I know issues we do like to quickly go over.

I

I

So so, just two small issues, so this one is for improving the locks, iron messages for the measures detail pattern apply command, so I have improved some messages here. So like this, you can see a restore trap could not create request for this one.

I

So these are pretty basic, but the problem I'm having here is some tests are failing, which I can't figure out this golang tests, which are type name, will be used as mesh model dot. Mesh model summary challenge. So if you could give some insight on this like, why is this spinning.

B

Internet give some feedback in the chat actually.

C

F

Yes, I I, provided the update on um what you can do to to fit that, so the uh go online doesn't like the the long line. What I'm sorry we we have a rule to check for the landline to be at 9 50 and the long line is, is uh I, think 11, 70 or something, and so I checked the uh the project looked at where it was calling it from, um and so it's on that particular file or whatever complaining about about the SVG files.

F

So once you just make the changes and then make it shorter, um that should take care of that issue. Now for this one, the the type name, let's see, I need to look at that one um I choose what there's something change in the mesh model.

D

Okay, yeah, this might have been introduced by uh I, think this mesh model summary Channel. This channel is added in pronounce PR when, via where we have the summary of all the mesh model components shown and this lint tissue might have been introduced there. So I'll take a look at this and I'll create a PR to fix this.

I

In your tongue, so it's not related to my Bluetooth right.

F

I

Okay, so I think if there's any, if you, if you can give any feedback on this and then you can give I've already requested a review from Aditya.

I

So now, moving on to my second item, uh just added some log messages in the engineic system, adapter and in these functions like resisting static, workloads with measuring server and successfully registered the static components with measuring server and then change this line at the bottom from successfully. It is successfully registered latest service mesh components with measure server to successfully register latest workload, components for version this, because I find I went through the code base for some other adapters and I found it like.

I

It was this instead of this, and this also like looks more beta appropriate.

I

So any feedback on this.

D

I, don't think workload. Components is necessarily the best thing uh they we used to use the word workload to reference. The thing when we had an open when we were using the ohm model in the new mesh model, we use components, usually mesh model has components and relationships, so in here I think successfully registered latest mesh model components would be the correct, like the more appropriate term. Do you have any Physicians over here foreign.

B

Just components, maybe is that I don't know if that's what you just said, but instead of I mean yeah. If you just drop workload and stay convinced and yeah.

I

Yeah, so I will change this from to mesh model components.

B

You know what or you could do what I just said, which is just drop workload and leave components, and then we won't have to talk about it again because yeah you just you just it's one. It's a common thing that a lot of Engineers fall into, which is over engineering like too much detail too much like you get yourself out of trouble by not being specific, so.

I

I

So like this is having two errors, so this file- uh there's a golang client error- has already pointed out by Antoine. Due to this.

I

This line in the config.gov, but I have already opened a separate pull request to externalize the config file, which moves all this data into a new machine, modeled metadata.json file, so and for those that these tests are passing golang saline. So we I really don't need I, don't think I need to change anything here or in this file, because this code, this code, are already being moved to a separate file.

I

So I'm not changing this, so this will keep showing the test Wing. Failing for this issue,.

A

So yeah yeah sorry to interrupt so yeah, so we are over time so yeah. Let's uh put this item and discuss it in slack. If you would.

I

A

Thank you for your update. I wish it was a good one, so we are three minutes over today and with that uh I think we are good to rap and yeah. There is a new promo. Video posted on play.message.io do go and take a look, I'll just paste, a quick Link in chat.

A

So, thank you all for joining today's message. Call we'll meet you next week same time by a.

H

B

Well, thank you girl, see you guys.