Layer5 Meshery Development, 2 Nov 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Meshery Dev Meeting (Nov 2nd, 2022)

Description

Meshery Development Meeting (November 2nd, 2022)

Agenda:
[Ashish] Enforcing provider #6440
[Ashish] Adding MeshModel components into the database.[Phase 1]
[Ashish] Registering MeshModel components on helm chart import.
[Ruturaj] Demo for workflow to update meshery.io/catalog
For publishing the applications publicly the user’s public design can be linked to their member profile.
[help wanted] UI for publish workflow
[Kamal] Implementing usage of Meshkit for logging instead of logrus/fmt #6350
[Harkirat] Meshery Catalog Publish From the UI #6478
[Kapil Bansal] Error messages in system lifecycle and system channel commands #6430 and #6437
[Rakshit/Gaurav] - Improve verbiage for toaster notification in configurations. + Next iteration in reference to notification framework.
[Pranav] Tombstoning MeshSync #6458 data and WIP #6435
[Nishant] Deprecate meshery/homebrew-tap for publishing mesheryctl.
PRs: meshery/homebrew-tap#32, meshery/meshery#6477
[Abhi] Meshery Playground[#6460]
[Abhi/Harkirat] Migrate from RJSF version3 to version 5 [#6379]
[Abhi] GraphQL Subscription Connection issue Fixes from client [#6244]
[Abhi/Gaurav] Need for e2e tests (contributor guide) of designs, applications and filters.
[Abhi] Refine and remove redundancy from the design/pattern YAML[#6369]
[Abhi] Fix problems on the Settings Page, the state views of Meshsync and operator states.
Todo: Fix the context-switcher too with the changes.
[Ashish/Abhi] Subscription refactor [Draft-#6409]
[Abhi/Gaurav] Need for tests for
Design Import
Design Delete (single/multiple)
Create Design
Workload validations
Catalog contents
Views change and rest

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

B

Welcome everyone to.

A

Messi's development meeting today is November 2.. So before we begin, I request everyone to please drop in your name in attendees list and before we start with and every each every every call. So we do welcome newcomers. So if you are joining this call for the first time feel free to unmute yourself and give a quick introduction.

A

So do we have newcomers today.

A

I guess not so maybe maybe.

C

I, don't remember um yeah, so so Jason Gomez is with us. That's nice Kapil is here, um Evie is here and uh botany is here.

C

Some of them might have been on this call in the past, but and a meat Amit Kumar is here just relatively fresh names, but some of you have been around for a little while, if you haven't had a chance to introduce on this call, um Now's the Time it's uh nice to get to know everybody nice to get you introduced to everyone else and everyone else introduced to you and that way they can help you navigate your way around.

C

The projects find a way to get involved, using the projects contributing to them or, however, um Jason have you gotten a chance to say hi on this call before.

C

Senor Gomez.

C

And we also have uh Papa dunda Ahmed joining we either have audio challenges or potentially some shy. Folks and I totally get that foreign.

C

If you haven't had a chance to say hi, um you know, feel free to drop in a message in the zoom chat or into slack, and we will be sure to harass you so or I mean get to know you.

C

um If you don't mind since I'm already um chat, you know talking a little bit um hey. My name is Lee by the way. For those that don't know me um and that's gaurav, that was just speaking.

C

um I just wanted to take a moment to talk about the currently uh um the release of messery, that's being developed currently, which is like the dot seven release.

C

So we released dot six a little while ago, and we've been going through a bit of planning for what's next, um there is a missing link that that um will get dropped in right around here it. The the link will take you to.

C

Oh just the the list of roadmap features for this. This current release that we're just we're just on the start of I wanted to take a brief moment to highlight a couple of items, um uh one being this concept that we talked about uh last time that we met. It is the notion that within meshery- and this could be an embarrassing moment- I just sat down on my machine and my running messery- yes survey says yeah. So what is the internet? What is that survey says?

C

Is that uh it's a terrible joke from uh survey says what is the Family Feud right? Yeah, that's the okay I guess you have to yeah um anyway.

C

um Survey says that I'm running meshery in meshery, when you perform operations like oh uh deploy, a design, um you know undeploy an application. This is a sample app by the way uh for a kind of interesting.

C

We've got a few different components, but if I were to deploy this and as a matter of fact just randomly, let's, let's do that. Let's see if my kubernetes is up.

A

Leah, are you shading something.

C

Man this whole time, I'm, not sharing uh I'm.

D

C

Oh, my goodness, okay good! Well, let me let me I. Can my demos are always better? Second time around? Okay, anyway, hey I, just wanted to okay. I didn't want to take that long. I just wanted to interject at the front of the meeting to say, there's a new release, Trend that we're starting a new you know, version of measury.7, it's uh gonna be packed full of features, just like every release is one of the bigger features is, was to be a set of enhancements around this area right here.

C

So when users, um like any of you, go through and interact with um measuring you have it perform actions like this.

C

This operation that I just invoked, it was an undeployed of an application that, apparently, is undefined, so I know gaurav is working in this area right now, um but anyway, what it said was: oh, it conveyed to me um what has transpired based on the buttons that I've clicked so I, clicked a button called undeploy and it meshely performed an operation that operation might be successful or it might not, or it might be successful, but take like an hour and conveying to the user.

C

um The fact that there was an operation performed how long it took if it's still in progress if it had a problem. What to do about that problem like retry, cancel the operation restart the operation, roll back. The operation I feel like we're on like, like I'm a weatherman right now, it's like okay, maybe I could roll it back. You might roll it forward. You might cancel it. You might even restart it and you, uh uh but you're going to want to know about all those things that transpired. So we have this Notification Center great.

C

This is helpful, but it is not nearly as helpful as it could be if it were enhanced to be more of an operation Center to let you know that the fact that you clicked this button at this time it you know and what happened to it.

C

So that's the first um significant kind of area of focus and enhancement in this upcoming release and there's a design specification for that area. It's going to require lots of front-end coding, lots of back-end coding and lots of user experience design.

C

There are user experience, design designs for meshery in this file called measure UI there is it's hard to see and I cannot blow it up, because that's not how figma works, um but it's there's a page called General, uh Notification, Center or there's a page called notification system, and it goes through and talks about how the current notification center works, but we need a notification center plus an operations center.

C

It does a lot more so so there's a lot of there will be any number of you that are contributing to the project that will be involved in this. Okay. Great, please spend some time go. Ask questions jump in let's get rolling on this. This is gonna, be good. There's some other areas that have already been in flight for a long time, so we're overhauling mastery's UI um I won't even speak to it. I'll. Just let you click on the links and come to get involved, lots of things to do there.

C

We've had we've called for operational reviews. I'm sorry, we've called for ux reviews previously um one this this so I'll. Let you look at that. I won't explain it. I've called four design. Spec reviews in other areas, and so I just wanted to highlight that these things are going on and that you are all welcome to come and express opinion or ask questions or say how silly the design is try to improve upon it and yeah.

C

These are just a couple of examples of things that will be happening here um if you were to gosh even just Google. This real, fast uh Zoom bars right in my way, I'll just drop this link in for everybody to go. Have a peek at this.

C

The ask here is that if you see something of Interest, please inquire about it, because there's a lot of question marks next to these things, and it's not the um if you're interested it's not that your name has to go here is that you might just want to collaborate on a particular area.

C

There's a fair bit going on. Oh who's. This person, all right boy, looks like they're gonna, be doing a lot in this release.

C

Okay, okay. That thing thanks: Garth I just wanted to uh yeah.

A

C

A

Yeah, so let me just give it give an overview of message also in panic. So if you are interested in this effort, please do what I mean in chat or in slack with that, let's move to our first agenda item for today, which is by Ashish. You have three agent items in rows, so Ashish would like to take over the screen.

E

E

Okay, so since uh uh what in one of those in one of these tasks, I was supposed to put a diagram into the architecture or the design spec, but since uh I was basically I, procrastinated and now I'll be showing it on the draw that I have to explain few things uh and I'll put that on the design Speculator so that mapled but uh okay, so I'll start with the enforcing provider PR. So recently, there was an Ask of having an environmental variable inside of machine server.

E

That will enable us to configure few things about how the endpoints are authenticated at the build time. So the authentication mechanisms were split into two uh provider authentication and no authentication. There were a few endpoints which require no authentication as adapters are the ones talking to them and they don't have. They are not provided specific, they do not have tokens and all so that's why they are kept.

E

That way, and other thing is that if you give it a flag which is enforced provider, then it will basically override whatever provider you try to log in as, and it will only allow you to perform operations via that provider. The changes inside of the extensibility API and extensibility provider docs have been made and this PR is up for review. uh So this and I had already demoed a little bit of this PR in the last call. So I won't go over this thing again and get to the next thing. I'll Circle back to this.

E

If anyone has any doubts, next thing was adding mesh model components into the database. Okay, so I I think uh yeah.

E

I'll first talk talk about this thing and then about next thing, so to get to give a little bit of context to people before I start mumbling about mesh model components in this I'll I have just created this diagram, and so and the green thing is basically uh what we will be getting to, and the red stuff is what we have right now, and uh so we have clients right now like match the UI Meshi cctl, and they make a request, and these requests are currently in a patent file format, and then these uh then, all of this, this big box, is inside of mystery server.

E

So this these requests are in pattern, file, format like one. You would see in maybe um if I go over service, much patterns repo, then all the currently everywhere you'll see you'll only find this format but we're trying to get away from this format. So this is how a general pattern file format looks like it has services and it looks like a Docker file and all so. This is the current format, and this is what client send request as right now.

E

So this is the final file format, and then we have definition files which you used to be definitions and schemas. So in pattern file format, you tell us the server what you want to deploy like your istio service mesh, your your infrastructure and these definition files are the way for server to know what these things actually mean. So the resistance or the handlers like they can be adapters. They can be kubernetes, it can be kubernetes itself. These things actually give uh it's like a custom. If you come from kubernetes, it's like a custom.

E

This is a custom resource definition and this is a custom resource itself. So that's kind of an analogy: I'll go over here, so these pattern files are basically a way for you to describe your infrastructure and these handlers uh resistor definitions, and currently they follow this ohm definition, ohm format which you we used to have and if you go inside of any of the adapters like if you go inside of istio adapter uh and then you you'll go inside uh templates home workload. We have a bunch of these.

E

So if I go into ohm and workloads, as you can see for each of the versions, We Are dynamically, generating these definitions and schema files, they are just a way for describing what it means for something to be an authorization policy artistio and when you get that thing how to handle that thing, etc, etc. So these are your definition files and they used to form, follow this home format.

E

And now we are moving to the component definition format and the instantiations used to be in the pattern file format and now we're moving to a design format. So the Handler's register. We use these definitions to uh to validate and to find find out the meaning of what uh like virtual service means and all and then using all that data we generate a request which used to go in the process home call, but that would be really improved now.

E

Process design call Anal using the analogy, but this is kind of the back end part of the back end. So even the pattern engine can be broken down into front end and back end. This is the uh this is how the end request looks like, and the way end request looks like will remain the same as of now, uh so the change is completely completely backwards compatible. So and after that a grpc call is made to the Handler. So whatever component said that hey I know what is mesh is or I know what traffic meshes.

E

That component will then take this request and process. It maybe deploy something, and then that will stream back events and, as Lee already talked about notification center, the message server would stream back notifications like this was deployed. Your mesh was deployed. This thing was applied in all those things. So now the migration, the way uh I have broken it down, is in phases.

E

So if I go over migration from ohm to mesh models, currently we are at phase one, which is that uh new endpoints are added under the slash API, slash mesh model, slash components endpoint and under these endpoints, new component capabilities would be registered. So as I told here that the resistant registers Things So. Currently it only registers ohm definitions in schemas. Now it will also receive component definitions in schemas.

E

This change would be in a way that uh there will be a time where patent engine will support both of these things and then we'll slowly get rid of ohm. We don't want to. You know, break anything so that we I, so we would be at a point supporting both patent file format and both the design format. We would be using both component definition and ohm. So if you give it a previous format will also support that and then slowly will deprecate it.

E

So, on the registration part, the ohm definitions and schemas will still be registered, but along with them, the first step is that component definitions will be registered and let me also open my pull request as a reference point.

E

uh Yeah, this is, and just to give you an idea about so initially. One of the other differences is that in ohm format we had separate definition file and separate schema file, and now we'll be, we won't be having it that way. There is just going to be one component definition file.

E

The schema is basically an open, API spec, so to speak, uh so every component, like STL, they are exposed via apis, and these apis have a way to tell the client that what are the fields and values that we expect or all those things come under schema and definition is all the other stuff like what is the name and who's the Handler. This come came from kubernetes.

E

This came from some metadata, like this came from artifact Hub, all those things they come under definition, so if I, uh okay, this is this is my PR and if I look at any one of these okay, so here I'm not statically, generating components but yeah uh if I load any of these up. These are the oh okay, I've deleted these ones, I guess these are uh yeah. There is the test.tml.

E

This is the one generated by nitish, so this is kind of how the component definition looks like uh it has the kind API version uh it has a schematic. uh Currently we are using this uvr putting this schema instead of spec, but slowly we'll move to uh schematic. This is for again backwards compatibility. We want to do a slow migration. Nothing should be very sudden, uh so yeah.

E

This is how the new format is going to look like, and these things will be registered along with these things that you saw but uh and how the way message server would work. Is it would uh it would base based upon what you? What is the thing? What is the format that you're trying to use if you're using designs it would use that format if you're using pattern files, it would use that format.

E

So the first step is at least to have the components at an end point for those components to be registered and endpoints for those components to be queried, and that is the job of this mesh model. Endpoint I have added the post the get and all the so there is post, get and I think delete. We didn't support, delete in the original endpoint, so uh I I, don't know whether we should support delete of components here. But again, that's a something of post discussion.

E

I'll come back to uh I'll, come back to the progress here, so yeah, so slash, API, slash model components has been. uh This API has been another thing that I just missed on is these definition. Files are used by server to create the instances? Yes, but there is another. uh There is another thing there is another place, also where these definitions are being used, and that is on client side.

E

So, if you're, using messy CTL, you would write your design files or you in the new format and in the previous format, your patent files by hand, so you'll type kind and all those things. But if you use something like design, configurator or something like in mesh map, you have designer. If you use those things, then you don't have to pre-fill those things. uh Why?

E

Because we directly go to the message server and we take these definition files and we make use of them to create the schema for you to create all the an rjsf form for you to fill so uh in the proposed model will be slowly will be using the component definitions to um basically the way it would work is the these things would be. So if you given a component definition, you can create a pattern. So there is, there will be a cross here and slowly from uh using a component definition to create a patent file.

E

We will go to using a component definition to create a design. So right now we use own definitions and schemas to.

C

Create a pattern, let me interrupt just because of the time check like okay.

E

C

Also, I don't know that this is more like probably a proposal for review um right, yeah I, don't think I know that we've talked about how to I. Don't know that this has been proposed in a doc and probably needs to be about. You know how to transition from one to.

E

The next yeah, all those things, are on phase two, that's why they were being discussed so coming back, I I was um I, was keeping from phase two anyways, so on phase one yeah, these endpoints would be added and the UI would use uh these component definitions to generate pen drives or whatever and uh yeah so and I'm.

E

Adding this thing in the same and the third task or the third agenda item is actually related to the second one, which is registering mesh model, components and Helen chart import so like if you import an SQL Helen chart- and uh you see, health chart consists of virtual service and destination Rule and you don't have them registered because uh currently we have an st adapter which comes and registers those components. If you don't have an adapter, we need to dynamically.

E

um We need to. We need to dynamically, have these components being registered. So let me demo that real, quick.

E

Okay, so here I have my mystery server instance running.

E

Okay and if I go over slash, API, slash model, slash components, I have no components right now and since I just opened the UI in a while I'll have here kubernetes components they are generated dynamically on starts yeah, so I don't have anything right now here, uh surely not anything like virtual service and destination rule I. Hope I'm checked out to the current Brands to make sure that everything works uh okay. So now, if I go and import, let's say an SQL chart.

E

Let me find a link for any of these, so if I take sorry yeah. So if I take click this one and if I try to import it.

E

To the applications.

D

E

Here my now, if you can see here this, there are custom resource definitions. We have like virtual service and all and as you had, and if I go to the components now whether there are a whole bunch of components, let me try to uh filter it. If I, let's say virtual server store.

E

Okay, it's I guess this would be the name for now yeah, so I have now my virtual service here, uh because and these this virtual service basically came from this Helm chart. So when you now import a health chart all the com, all the custom resource definitions it has, it will use those custom resource definitions to refer back to this thing, the component definitions and this endpoint. You can use to see those component definitions in the scheme and everything so yeah that change has been added onto the uh onto the same PR yeah.

E

So that was it from my side and this uh the phase one that I'm talking about this would be this would be over by um by this week. uh If I get some support from Front, End um front-end will have to use these specs basically to slowly migrate and I.

E

Think the first step would be to use kubernetes components, um as one of the other tasks was to have these communities components on your local file system, so that, even if you don't have a kubernetes cluster connected, you can still design your service and just you won't be able to deploy them. But you would still be able to design them and uh for that that change is also added in this PR I'll go over time, so uh I have already demoed it uh another call so I'll just end with that note. Thank you.

A

Yeah, thank you Ashish, so we have next uh rituraj. So would you like to present on YouTube.

D

D

So, like literally required, um the the UI for the publishing and the in the meshery, so I added it and like like this, is all like stuff. So if anyone who is admin can like test, my PR like to tell me whether this is actually sending the payload like it's actually sending in it like I, have don't have the admin role and just confirm it like it's actually working. So that's the update regarding adding adding a publish uh button in the UI for the mystery. Like a small update, yeah, you can you see the pr.

A

Yeah yeah, if you go back to UI, so are.

F

A

The same problem component yeah.

D

uh No, this I created a new problem.

A

I I think you should use the same problem component.

D

Okay, uh do you like one this one I uh like I, think this one yeah foreign.

A

Yeah: okay: we can carry forward the conversation in slack, yeah yeah. So next up we have uh Kamal on implementation usage of mesh kit, Kamal IO for demo.

G

Yeah so uh I hope my screen is visible.

A

G

uh Okay, so the my PR like this is regarding uh for login like currently, we use a logger or fmt package, but they are the external packages that we use for login, but this PR is regarding now now that we have our own function, Library called meshkit, so we would like to use its uh utis.log for logging, so this PR.

G

Currently it has some uh unit test errors, like one error, is like um the Import Cycle one second yeah, so Import Cycle not allowed so yeah I was just going through uh those errors and trying to fix them.

G

Yeah that was my update.

G

I would fix them quickly.

A

Thank you Kamal. So do we have any feedback for Kamal.

A

So yeah, let's move on to next arranger item, so we have a couple. So Kapil would like to present on error message system, yeah, sure.

H

H

Is my skin visible? Yes, it.

A

H

Yeah okay, so my task was to improve error messages in some commands. Like first of all, one was messy system Channel command. There I tried to improve messages when wrong argument is passed, no argument is passed and more than the required arguments are passed.

H

So I just want some reviews on this. Piers like there are two PR open one for Mastery system, Channel and other for system lifecycle commands.

C

Yeah I I got to take a look at one of your PR's and um I. Think Kamal had one here as well. um Fantastic I, like that. Those are those are really helpful, make a huge difference to the user as they go through and try to you know they come to a CLI for the first time and they type a command.

C

Of course, if they're they're going to get it wrong or I think there were a couple of times recently I was just I was demoing, mystery CTL and it had been a while since I'd used a particular area, and so I was just looking for help. I was um using the help feature to get some guidance and I wasn't getting the best of guidance, so I mean so the these updates have a big impact. These small updates have a big impact. It's good.

H

Like I, don't know whether I covered of that task completely or not, but I tried my best, so it would be great if I can get some reviews on them.

G

Actually I saw I saw this PR. It is really great.

G

F

You, like I, saw the changes.

H

Although I think like this lint test is failing in my peers, it's.

A

Failing in master so we'll get them fixed.

H

C

Yeah, having said that, um there's actually we could use this opportunity to explore a little bit because one you might, even if you didn't introduce the lint check failure, you might just fix it anyway, either in this PR or a subsequent one, and this also for those that are on the call that are kind of looking for um low hanging fruit like an easy place to jump in. um This is one of those places and it's really helpful again. I, don't know how familiar all of you are with sort of uh wading through the notes.

C

But if you click the details button to the right of that failed, lint check, it can take some finesse. But if you yeah, oh geez, yeah, you've been through here, perfect yeah.

H

I would like to look into this.

C

H

Open a new PR for it.

C

Yeah, please yeah it's um this one's a little bit um different than the typical failed lint check, usually the failed lint check will say uh it'll, say any number of things, but it'll talk about your code and what's what the issue with the code is? In this case it looks like it's saying it's having some problems, something.

H

Can't load fmt cool.

C

I'll leave it in your capable hands, but.

E

A

Yeah, thank you Google. That was a good update. So next item is by rakshit and me, so it it's on improved I.

H

Also want to discuss like we have labels like I asked on slack and they said that label is updated, but the label accepted by hectober is accepted.

H

So this label is wrong.

C

Yeah do they um need the hector Professor is listed as a topic in the repo do. They also subsequently need um labels on each individual PR.

C

C

Yeah I don't know that the I guess what would I don't know that the label matters uh like it um I'll point out the rules and yeah. Maybe it's because the pier yeah anyway.

H

C

Yeah, let's follow up. Let's.

D

C

Get that credit.

A

So yeah moving on to next topic, so it's by me and rakshit. So it's on improved bh4, toaster notification and configuration and uh the next iteration for the notification framework in this. In with respect to the notification that we have so earlier. In this call, we saw that we were getting an undefined on this thing over here in the toaster notification. So it will show the item name and uh notification, showing design successfully deployed and similarly for deploy and undeploy as well and the same thing is updated across patterns filters in both grid and table.

A

A

So I collaborated with her attracted on his PR and the next item we would we want to work on. This is with respect to the notification centers So. Currently, we are just showing a green success, snack bar on deploy and undeploy, but we want to change it to uh blue information info snack bar and because we will be adding using the notification system. So to make the consistency, we will be changing the success bar to the info bar.

C

It's not a question uh question I, I, don't know like and Ashish might know this, but by the way just I was just noticing.

C

um Most people don't have their webcams on, like hey. If you've got a webcam turn it on it's great to see, you um I have a question about notification center from the UI. If you would like to write JavaScript to enter, to create an entry in the notification center, is that possible.

C

Or do all of those need to be sourced from the back end from golang and.

E

C

E

Are you completing it well,.

C

Okay, yeah, even if they do need to be sourced from golang like from the back end, is it possible to is there a generic mechanism for the front end to say? Oh, hey, I have a new message I'd like to send this text. You know, maybe maybe the you invoke a back-end operation that would then you know, you know, enter it into the notification center.

E

Yeah, yes, yes, yeah.

H

E

So uh actually, actually uh I mean whatever is seen on. The notification center would be the result of what is the state of the notification and that notifications state would be structured in such a way that it is robust and maintainable in a long term that it's follows some uh structure of what we get from the back end and definitely what you get from the backend from the uh by the updates in slash API, slash events.

E

uh It will be truncated as a next notification message inside the notification panel and if it needs some more uh structural supervision or structural change, modifications Etc to make it more uh I mean make it more generic and uh more easy to handle in a long term. So it can be done because that's really an old implementation of what we are seeing right now. So definitely we can do any number of things. We can trigger an event. I mean notification even from the UI.

E

Also it can also accept some thing from the back end and because that's all being streamlined through a react state, it can be done. Anyways.

C

Oh, that's great.

A

Yeah, so any other feedback or question regarding in this particular item.

C

So one other one: if I could sorry one other one if I could I think you know, given you know how frequently the notification, center and sort of the enhanced kind of operations center will be used and referenced by both people contributing and users.

C

um Well, I, guess mainly by contributors, it'd be really nice. If the set of contributing docs that we have in the docs.mestery.io had a little description, uh you know for UI contributors about you know what what to do and maybe I understand that might change a bit a little bit over time, but but at least we could document um what's there now and it could be a great reference.

C

um You know from going forward, yeah.

A

C

Thank you, nice.

A

Yeah, so with that, let's move to next agenda item by pranav, so pranav, would you like to present.

F

Yeah I'll share my screen.

F

So I will be talking about the tombstoning of items which is assigned to me right now. um So if you don't know what what happens when next, we when Messi is deployed where the inside dots are cluster and it discovers a cluster, then it um automatically identify the workflows running in the cluster and then it shows them onto your dashboard and currently what we have is that if we delete uh any of the items and one more thing that it saves them into the locals um local database like right now we can see all them.

F

You can see them through different extensions, So, currently you're asking that if we delete any of the workloads then, instead of deleting them from database, we should Tombstone them so that we can refer them again, but they shouldn't be shown on the UI. So that is the thing which I'm currently working on, so it's not completed, but I can show you demo like what it's what is happening right now.

F

So, for example, if I created a namespace test um test, one which I just created right now, and we can see that it's in this table and if I delete it right now. So even though the UI will get updated based on if I do.

F

um So when this item is getting deleted from my cluster, it automatically reconciles with my sync and messaging sensor, a new event that is deleted and, as you can see, it just got reduced to five and similarly five now so that in my database, I wouldn't see it because it automatically deleted as well as soon as I. Do that so now, I will be working through that through a soft lead, a new tag which I would be adding into the models which we have.

F

We would be needing one more that is deleted at field which will be tracking. um Whenever we delete a new any of these items, then, instead of deleting it, it will add that um that time, step one which will get deleted and then we can track them on on future reference as well. So that is the thing which I'm working on right now, so any questions or anything which I need to consider.

F

C

Yeah I wonder if I wonder if it was there caught that I have to say I missed part of what prophet said, although now I'm very sorry for that, because because it's a great feature that you're working on um to replay part of that is that ifs related to this functionality. If someone disconnects a context, if someone removes a kubernetes context from mashery that will that has the effect of removing the meshery operator from that cluster and I guess this is a question.

C

Consequently, does that also have the effect of tombstoning all the records associated with that cluster.

F

Currently, it is not in place, but after my period would be currently even on deleting a context. We don't delete the matching data which is in the database, so I already have a PR. I was integrating the new thing as well in this one. So it's not completed right now this one though it takes care of deleting the data as soon as you delete the context.

C

Okay, got you got you um and then in the that would be similar, Behavior will occur if a user goes over and chooses to they leave the cluster connected. They leave, um but they decide to flush. The mesh sync data is that the same will the same underlying system Behavior occur or will it be different.

F

Actually, currently, the thing is that if you flush our matching data, then we also do a leasing with the connected cluster. So that would require a little bit of thought on how we want to change the current behavior, because if you don't want to read using our flashing the data, then that can also be achieved. But that would change the current behavior. What we have right now so yeah.

C

Yeah got you yeah, it seems like a simple initial answer would be to hard delete on flush, given that a resync will occur um and I think with mastery's current functionality will that will generally be acceptable? Fine in there there will be a future consideration where people are admitting their clusters and something happens. You know and they want to go back in time like it happened yesterday, and they want to go back in time to look at that, and so.

A

Yeah, thank you. So next up is Nishant. So Nishant would like to present.

B

uh So uh in the last call, I told that I'll I'll be making a producers to deprecate our custom home brew tab. uh So for those who don't know earlier, we used to use this tap. We still use this tab for publishing mystery, CTL uh binary so that users can tap. uh Users can use this command.

B

Add this repository to their home View, and then they can install install Mastery CTL, but they no longer need to do so because mystery CTL is also present on Homebrew core uh and therefore we and since now we are publishing our releases just fine uh from using Homebrew core. We are deprecating this tab and for this, I have made a pull request.

B

uh This pull request deletes the sdctl formula from this tab and also adds your tab, migrations, dot, Json, and this file is understood by uh Homebrew and uh Homebrew knows that uh Now message serial has been migrated to Homebrew core, so anyone who has installed a mystery CTL using this tab will now receive updates from home review code.

B

App and I've also made a second pull request that uh that modifies the go, releaser and uh and we'll uh so whenever the mystery CTL release workflow occurs, it will no longer try to publish a binary to publish like update the home group formula in our custom, tab and yeah.

B

This is pretty much it and the only uh and like as I, also told that there is- or there was one uh consideration- that we need to keep in mind uh for QJ while publishing to Homebrew, or is that so uh in so like previously, our religious release.

B

Titles uh went too long release versions so, for example, 0.6.0 hyphen rc.6ff, and these uh right, these versions were not very uh like were not being understood by Homebrew and therefore we had a so we we had to define a version function instead inside our Homebrew formula and that correctly passes the version using our using the branch name and uh but as soon as our version titles got simpler. Like 0.6.2 and Homebrew core was able to pass these versions.

B

uh The the sci start started erroring uh saying that you, you know no longer need a custom version function, uh so some so one of their maintainers removed the version version field from our formula and so that it can be automatically detected from now on uh but like if, in future, our version titles again become complex, like this, we'll need to uh manually modify the uh formula again in airport request to Home Depot.

C

Yeah Nishant there's, no doubt we will run into this again, because the version numbers the versions that we were using, even though Long they're correct their sember compliant um and it's homebrews uh lack of sophistication, that's getting in our way um somewhere, so that we can.

B

Go back to begin and another permanent fix would be uh like we can make like we're, using this uh action to publish our uh to public to make pull requests to the Home Group repository. As soon as we run the mystery CTL release flow. uh Maybe uh we can modify it uh so that we we can modify it so.

D

B

B

Yeah uh we can modify it so so that, if, if our versions get too long, it automatically adds the version field again or maybe make a pull request to it, and.

C

It's that sounds reasonable. You know.

C

The first thing you said sounded kind of nice.

B

uh As in like forking, this and making a change.

C

Well or as in like or I thought, you were alluding to potentially having logic in our workflow that checked, for that it works around the lack of sophistication of Homebrew by.

B

Yeah yeah, actually the complete workflow, uh the complete flow of making a pull request uh like editing. The formula is handled by this action completely so to make modifications that will need to make changes in this action either by forking it or.

C

Oh yeah sure yep.

B

So so that we never need to worry about this.

C

B

uh So, like should I open a GitHub issue tracking this.

C

Please, if you would yeah.

B

A

F

C

A

C

Yeah I think I think we explained this last time but nishant's. um What he's concluding is that the installation of Mastery CTL is even brief or even easier than it was in the past and so he's just tying off that work. So so it's quite.

D

A

So next we have Abby with multiple issues in a row, so Abby would like to present.

E

Okay, yeah. Let me share my screen.

E

So I hope Chrome is visible. Name is uh uh the first issue is the Machinery playground, so we are planning a new hosted environment of Machinery that could be used by uh anyone, and for this we need limited feature sets and uh based on the user. If it is so based on the user, I mean I. uh If the user isn't I mean not, a user can be entitled to some some features and they cannot be entitled to some others. uh So this is the work under progress.

E

Where you know the entitlement will reflect on the UI, which part of it should be visible to the user and which part of I mean which part of it user can. You know, interact with and and whatever, which are which of them. They cannot So. Based on this, uh there may be any number of uh non-pointer events on the UI and based on. What's uh what you are entitled to you may have, uh you may have the certain part of the UI functional for you and certain part of not so so, I mean yeah.

E

So that's that's the that's item number one and since I have less time so I'll be moving to the item. Number two, if you have any doubt, you can ask me, are moving to item number two. There has been uh recently uh rjsf. What we do is uh so this thing is also uh I mean the.

E

Since my capabilities uh I mean my capabilities are saying that I can't use this, so I can't show you the configuration, but the same thing is available here also, so this is the form that we use to, uh for uh you know to configure the I mean fields that is available to the user, that is to be included in the final EML of for the patterns or so-called I mean this is not called design. So this is the RSA form and uh there has been.

E

We were using rjsf version 3 for a very long time and we have just moved towards the rgc version.

E

5 and that's that's somehow, uh I mean yeah handling, this migration change and we have a few regression issues, but her girat is on them to fix it so and moving to the third item, which is graphql, subscription connection issues uh recently I mean what we had earlier was uh there is a subscription based update that we do uh in the UI whenever the Whenever there is some change in the database or sorry in the mystery server I mean in the database.

E

Machine servers tells that there is a change in the database and that triggers an event, and that shows like uh the result. So what's been changed or not so earlier, the uh UI was I mean UI was the cause of a big quarter, link that we have around for a very long time that the connections get lost after many retries. If you're seeing here, there has been more than 10 to 15 retries and all the connections have been lost and the final connection is uh established.

E

So based on I mean we were for a long time.

E

Thinking about this is a bottleneck from the back end, but uh it was not and came out that there has been some issues around UI where the subscription I mean the connection time uh for the server to accept uh I mean to acknowledge the websocket is very less that it drops, so it just starts cropping simultaneously I mean regularly in Progressive Manner and then the final one of them just get accepted, and so that's being fixed fixed with increasing the timeout minimum timeout and I mean that has reduced our uh connection.

E

Speed, like initial connections, uh connection, speed to uh I think from 10 15 seconds uh to like one two seconds. So that's been a good performance update here and uh I. Think I have covered up to this point need for e2e test. So there has been a long standing ask for I mean ask for the villain volunteers to write some con uh end-to-end tests for designs, applications and filters.

E

So again, uh I'm unfortunate that I cannot show you the design, application and patterns right now, but uh what they are I mean most of the people on the call were signifying like they are the part of the configuration menu here and what it does is like. There has been some initial set of tests already written for design, applications and filters, and there has been some updates around I mean there has been movement from a restful API to graphql and uh that broke some part of the test inside uh there.

E

So those things needs to be fixed promptly in order to you know, move further with adding more of the end-to-end test. Support for measure UI with respect to these three constructs that we have and that's why the volunteer is needed here and I. Think Mario also has uh outstanding PR here, I, don't recall like whether it is linked or not, but I'll see and yeah. We can see through it all together to rectify the problem here and even have anyone has any interest over writing YouTubers. Then they can uh Tell It On slack.

E

There has been two channels Associated for uh test. Only that is one is missionary. Ci and another is Cyprus testing. They can show their interest anywhere.

E

The say next item is refine and remove redundancy from a design file, and that's related to uh this, where there has been bunch of duplicate information uh going along where I mean I, it may be a refined ml, but what you are you you used to see here is the name name: space and labels. Labels and annotations were mentioned.

E

It two places that were in the first of all here from line 16 to 20 and second as a part of the settings, and they were both redundant to each other and they were for no reason increasing the final pattern file size so that related redundancy or that duplicacy has been removed in this sphere. So that's mostly related to it.

C

E

Yeah next is so that's that's that's a very long. I mean very. What should I call very a past or I mean I can't get a word, but I I have no context right now for this, because there has been uh increasing number of problems in the settings page and uh messaging, an operator States. So there is one of the this uh thing where the subscription refractor is one of the part where we are doing some changes uh with respect to refactoring.

E

The subscription and operator status uh to you know, make things uniform uh from the front-end as well as backend, and uh I mean these two uh things can be, you know jotted down together and what what? Basically it's telling is that we were into issues with the mesh sync and operator state for a very long time, and we are doing some updates regarding subscription uh for the operator. State and unifying I mean mesh sync and operator.

E

States right now has been uh two different subscriptions for a very long time and they are being now unified, uh and that will help us. You know to drive the consistency and uniformity across the UI and handling less education in order in the UI to get into any future works.

E

And yes, uh the last item here is is related to the end-to-end test. For uh here only which is I mean there has been I mean. Whatever you see in the designs, uh I mean these are the list of tests that can be that are very beginner friendly. To do I mean that are very beginner friendly to write and they would all use they are all using API requests and no graphql.

E

So that's why they are beginner friendly, and there is a need and ask for volunteers in this area since I mean writing these tests down one by one or any one of them by anyone. What's what time yeah so uh this will, you know Aid a it into your current knowledge of machine UI as well as, and this will be a good support for meshri. Also uh in the form of you know, extra end-to-end tests, I think that I have been talking.

E

I have talked about this in this uh bullet point only I think cam over.

A

Yeah, thank you, Obi. That was great set of updates from you, so since we are over by three minutes, so let's call this a wrap. So thank you. Everyone for joining today's Michigan fall. We will meet next week with some new and more agenda items.

A

Thank you for joining everyone. Thank you. Thank you. Bye.