Layer5 Meshery Development, 2 Feb 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Meshery Development Meeting (Feb 2nd, 2022)

Description

Meshery Development Meeting - February 2nd, 2022

Agenda:
- Demo of import feature in patterns.(#5038)
- Compatibility Matrix Updates #5009
- Switched IsMesheryRunning with AreMesheryComponentsRunning. #5020
- Configurator feature + style enhancements. #4954
- Design updates
- WIP Github action to apply pattern #25
- Mesheryctl `stop` waits for delete to complete (#5021)
- Call for volunteer: Design/UX - Service Mesh Catalog on meshery.io - #658
- Call for volunteer: Design/UX - Pattern Catalog in Meshery UI

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

So, hey guys welcome to the machine development call. We are three minutes after so I guess we can get started uh so since we don't have any newcomers on the call. Let's get started right off the agenda ashish. Would you like to please take us forward.

B

Can you see my scene yep, yeah, okay, so ma so I have. My task is to demo this I've added this import feature in pattern. So patterns are evaluated in stages, so there are various stages in patterns. uh I have uh added uh one more stage onto those which gives it the capability to uh use external patterns. So, for example, right now we don't have this capability that I'm showing right now, so the types can be that this is currently not a valid type, but uh my pr uh what it does is.

B

Basically there is um what the first stage does is. It goes into each of these services and if there is an external patent file attached to any other service, it goes to this particular service and it takes all these uh services from this pattern file and then expands that to here. So and it does it recursively. So you can have patterns inside of patterns inside of patterns instead of pattern.

B

So basically it gives us a lot of composibility, for example, right now here you can see a pattern of 21 line and when it gets expanded after the import stage it's around.

B

This is the expanded pattern which is around how much 294 lines, so it will allow us to get away from huge and large patterns, and we don't have to scroll that much. The idea is that you can um any patent is an exportable pattern, but a good. uh Basically, if you want to make it externally. Configurable you'll have to supply this vars field, and these are the default values for these variables and these variables can be set by the calling pattern.

B

So, for example, I'm calling this uh istio4.yaml and if I go into four dot yaml, if I'm going to use twofold.tml, then it uh exports these fields, so you can configure control, plane, namespace your database namespace, here's the version and profile so recently. I've also added a way to because istio supports three profiles: demo and then there is default and minimal. So, depending upon you want the ingress gateway or egress gateway, or just the control plane, so yeah. So, for example, in this particular pattern.

B

What I'm doing is I'm asking it to go and fetch this seo. Installation pattern expand it here with these configuration, uh so the control, plane, name, space and sidecar namespace is keywords and everything would be configured inside of this externalized pattern, and then I'm asking it to wait for this installation after that is done, install my add-on. So by this we can have add-ons in another pattern file. So this is another pattern file which has grafana and prometheus add-ons uh I've.

B

I haven't made it a lot of configurable because it was only for demo purposes we can have. um Basically, we can have a lot more configurable fields right now. I've only made the control train namespace as a configurable field, so that I have set here a system system and uh the uh after the services are deployed.

B

I also want my booking for app to be deployed, so instead of rewriting the entire book info up here, I've just what I've did is I've just referenced this other book info app and uh with this book info app externalizes, basically exports this namespace field, so you don't hard code it, and in this case I'm saying, install my book info app in the default namespace and do it after my installation is complete.

B

So basically in our 21-line patent file, I'm describing it to install my service mesh with proper configuration, install the add-ons and then install my bucket for app so time to demo. So this is my current uh current running pods.

B

As you can see, there are nothing there is uh no booking for anything and uh if I go into my easter installation, so the the idea is that each pattern file each patent file that we import from somewhere else should be com should be uh complete in itself, so that when we will I'll talk about it after I deploy it so that while it is deploying, I can talk about it. So, for example, here I'm also enabling site automatic cycle injection in the sidecar namespace, which we have set as default.

B

So when I'll deploy this pattern, each of the pod in the book info app will have the onboard proxy by default. So if I do pattern apply, I okay and if I go over here and if I wait few seconds- okay, so yeah while that is uh taking place. uh So the idea is that we can have a service mesh catalog and instead of the service mesh catalog, we can have a lot of patent files which are which act like an atomic unit in themselves.

B

So a complete pattern file where we can specify uh first a prerequisite for this pattern file. For example, if I go over the book info patent file, you can see that it has. It has things like virtual service and destination rules. So if we basically take this pattern file- and we put it in the catalog- we'll say that okay, you need a service mesh installation because you there is.

B

There are things like virtual service and destination rules here and uh and so first requirements, and then the other thing name space everything that you can configure in this pattern file. In this case it's just namespace. We can add more things here and you want these to installation. Well, we have another pattern file which does just that, and these are the variables that you can configure. So basically um user can go to the catalog.

B

Add a bunch of uh you know just keep keep copying the links, add a bunch of stuff and on a 20 line pattern file. If I go over yeah, so as you can see my istio system in my ht system name space, I did a minimal installation, as you can see in the configuration so that yeah the minimal in the minimal uh installation only the control pane is installed.

B

uh So only my code is installed other than that my add-ons are running and in the default namespace my book info app is installed with my online proxies so yeah that that's the extra thing that I've added these are just the logs from the pattern apply command.

B

Let me delete those so yeah that was the that was the demo from my side. If anyone has any questions regarding anything here or you can ask.

B

Basically, the advantage of one of the advantages is that we are trying to get to a stage where we have what we call uh higher higher order components. So basically, for example, you can have a circuit breaker component, uh and that is agnostic of the service mesh, so you can have a so you just describe the configuration of your circuit breaker. You can apply it onto any of the service mesh and internally we would handle it, but by the time we get there.

B

What we can do right now is we can have our own circuit breaker patterns for each of the service meshes. We can have one for traffic, one for link id one for stu and everything and we can export these variables and all users have to do- is just take this, but take for example.

B

In that case, it would be something like um you know, the url of that particular url of circuit, breaker, istio and then uh will, and there would be uniformity in uh that in each of these circuit breaker patterns, like circuit breaker, stereo circuit, breaker, link received different circuitry patterns will have the same exported field so that to user uh they don't have to care about what internally that patent file is doing. They can just configure that they're using these uh arguments.

B

So it's like calling a function with an argument and that the definition of the function is the pattern file itself uh so yeah. That was uh that's kind of the whole summary of it. Anyone has any questions about it.

A

Guys any questions or feedback.

A

Okay, I guess not, then uh thank you ashes. That was a great demo, so uh moving on to the next topic, uh aditya, with your compatibility matrix, would you like.

B

To share your screen.

A

Yes I'll share my screen.

A

So yeah so as a suggested in the last website meeting, uh we are to be uh right now we are showing both edge and the uh latest edge version, as well as the latest release. So how I made this work is like. If there is a latest release, then uh the release uh the edge version just prior to the release, then that that will be the one which will be shown in the first home page and I have uh implemented a filter kind of thing, which at least suggested.

A

I don't know if this is feedback open. If this uh a desired change or not. uh So if we go and see uh past results, we can filter according to edge test or stable test. Like we select edge test, we will only get edge test and if we select stable test, we'll just get this table test uh of the selected adapter.

A

So yeah, that's where the changes I made from the.

A

So one thing that I would like to add is maybe not have an empty table when they land on this page. Okay, we can go.

C

Ahead with the.

A

Filters but like the empty state, should be all the results like edge and stable together and then.

D

E

A

D

Okay, so when user clicks, the password button just display all the results.

A

Yeah, instead of having none, that's all and then filter is a good test, but that's probably not the right placement for that yeah yeah, just I I wanted this for demo purpose. So like uh is do we want a check box or like like? Can you suggest where should this be yeah, I think so. We can only select one at a time right, yeah we can select. No, we can select both yeah. We can select.

D

A

Yeah, it shows so if we have an empty state of all the uh test results, so probably a radio button would uh would be a better choice that we could only select one at a time. That would be.

F

A

Right, okay, so if you are showing the uh past results of initial state as all of them, then having a radio button will be yeah.

E

Yeah all right.

A

Any anybody else, any feedback in the comments.

E

Yeah in terms of placement of the checkbox or the radio button, you can have it at the right most side, yeah.

A

A

Yeah, so now uh all right. So if that's the thing now, the only thing that's remaining is uh the breadcrumb issue like if anyone has like. I have not looked it into too deeply, but I'm not able to figure out what the problem is or how to set this breadcrumbs.

A

If anyone has any feedback or like help, I need help in this thing, like I'm, not able to figure out how this thing these breadcrumbs work and how they redirect like. What I want is that when someone clicks a past results, uh we should have a field here like slash and ht or past visuals kind of thing, but that is not working.

A

I'll, probably let god.

E

uh Yeah sure so you are seeing the functionality of for the betterment, so the best example would be the pattern configurator that we are using. So it also has the breadcrumbs and the functionality you are looking to implement over here is like already implemented there. So you can refer to that similar one.

A

E

You were also saying you can add. Yes,.

A

So basically the breadcrumbs come from the url, so you uh you created a folder called stu and then that has the uh test results right, uh yeah, so yeah under that message: yeah yeah! So that's why, since you see that uh you see the.

E

A

This is the I've. This is the url, uh maybe yeah. I've.

C

A

Base url, okay, you could take a look, there's a includes called breadcrumbs.html. uh That has a logic and you can take a look yeah yeah I'll look into that and ashish does. Does that sound fine to have the last edge and the latest results uh knock it down. Yeah.

A

All right- and I think I think the linkered uh like 5.20 is- has not been uploaded to the docs like it. The test was failing for linked 5.20.

B

How much older this version is? When was the test ramp? It was thirty four. uh I think we would need to get rid of few older uh few older uh test results because, for example, in linkedin things were failing before and those are wrong results or stay. You cannot call them states, they're, basically false alarms. So a lot of these red things are just false alarms and they were failing because of some other issue and not because mischievous, not com, able to support the installation so we'll need to uh you.

B

Can you can pick me later I'll? Tell you which of these to delete? Okay.

A

A

All right, that's it then.

A

All right, thank you. uh Moving on, would you like to share your stream yeah sure.

G

Okay, that exactly I am okay, nice, let's start with this residue so right now I want to see my status of mystery components, guess what it is saying that message is not running, but if I see with cube ctrl, there are some parts running and I can't see the status.

G

Similarly, if I like to stop them.

G

It says machine is not running, but I want to get rid of these pods. How do I do it? So that is the issue and that we are talking about right now. Messy studios should be smart enough to know that there and there are some deployments, uh even though measuring is not running here.

G

I can't see measuring here, and there are other adapters and mystery broker is here, so it should be smart enough to know that these are the deployments running and the they need to shut off okay. So what I made the changes is now. uh Let me okay. So previously it was like this was being done with, like is measuring. We were trying to check if there is a pod that is known as measuring is, is it running? Is it active or not, and then we were executing these uh commands?

G

Like start sorry, stop and status or log in all those uh commands, but these should be what these should be. Actually checking is if there are any measuring components running so that's what it has uh like changed so right now, if I'll, let me go to that.

G

Demonstrate it right now.

G

G

Okay, it is taking time.

A

Yeah, I'm in the right branch.

G

I'm not sure why it is taking time right now.

G

Okay, yeah nice, so if I run system status so even though machine is not running, I can see the status of other parts other deployments. And similarly, if I want to stop these deployments, I want to clean up my cluster I'll. Just say that stop it, so it will uninstall these running deployments.

G

It was a issue faced by a lot of people when, like the age of these clusters are like a very high and the deployment like the pod is not running and not active now, so it will solve that problem and also there was one more problem that that has been like faced by a lot of people that is like whenever someone is trying to install it, they face issues like some errors like this uh cannot start measuring. There are something already existing unable to continue installation right now.

G

Mystery city is uh like implementing a helm based installation. Sometimes it doesn't work because because of uh irregularities, when someone is like, if someone ran system stop, but it didn't, it didn't cleaned up cluster very well, like it left some resources there and next time. If someone will try to install it, then it will cause errors like that that can't install it, because some something is already existing.

G

So to like clean up that a new flag has been introduced. Let me show system stock dash. It's called dash dash force, so uninstall machine resources forcefully what it will do right now.

G

Okay, so right now, what will happen is with the flag force of dash dash force. What it will do is right now it will grab all the deployments. It will uninstall all these deployments. It will get the services, it will uninstall all the services it will get the replica sets, it will delete them. It will get the stateful sets. It will delete all of them. The custodial bindings cluster rule interface, cluster rules, every resource that has measure in it will get cleaned up forcefully via this flag.

G

Okay, sorry uh for that noise, uh someone came behind me, okay, so uh if you have any doubts and if I fail to explain what I'm like, I have introduced with the flag, so let me know.

F

So please one question: so the order in which we delete the resources is fine right like uh like. If there are dependencies, so the ones without any dependencies should be deleted. First and like, for example, deployments and services right, like the ordering in which they get deleted, should be fine right.

G

Okay, uh yeah, that's a good question and it is. It is okay. I have tested it like if I start measuring again and then uninstall it with forcefully the force flag and then restart measuring again. It will work completely fine because there will be no existing uh existing uh resources there and, uh like the both implementation, the both installation and installation process, the helm, one the hand based installation and the forcefully uninstalled uh the process. They don't collide with each other, uh so it will be like a.

G

They are isolated like they are implementing similar but they're different implementations of uninstalling and installing.

F

No, my concern is like uh when you are trying to delete, uh let's say deployment.

H

F

It shouldn't fail right in this code. The delete should not fail for saying that it is being uh this particular deployment is part of another uh kubernetes resource, which is active kind of thing. So that was my question. Okay,.

G

Yeah, so I am only deleting all those like every resource of measuring ecosystem has the word measure in it, so these deployments will only get deleted if they are from measuring namespace, uh whatever the namespace is. The guy, like the user, is using that namespace and if it has the measuring keyword like the name in it only then it will be deleted.

G

And it is for the like only for those cases when, uh like user is having hard times because of those problems like already existing thing or if helm is failing to update the images inside the deployments like that also happened very recently, when some big changes were introduced, it wasn't working well. So in those cases this flag will come in handy.

A

Anyone else any feedback or comments on this.

A

All right uh thanks, pierce thanks susanna, so uh moving on uh gaurav uh yeah. Let's.

E

A

Your pr on the configuration.

E

Sure I'm sharing my screen.

E

Yep, so this pr consists of two issues: one is for having integer type as of num for the text field that we have and converting the additional properties to the properties that we were using for the basic rjsf and moving to the basic uh custom component that we are using over rjsf. So, let's demo that thing so previously, what I did was I directly imposed the condition of maximum and minimum range to the ui custom component that we were using.

E

So what issue it was um raising that it was changing all the text field that was having a type, a string array or a number and integer, but we only want the range to be maximum and minimum for the type as for number and integers. So I added the type um type configuration in the schema so that it won't affect all others properties. So when we type something over here, so it will say to the user that it should be a number and the range is set to for zero two hundred for replicas.

E

Only so if I added one two three four five six and um it will indicate the use that it should be less than or equal to hundred, so it is not affecting on this kind of properties and all strings.

E

So next update uh next issue covering in this pr is about um having labels and annotations as additional properties. So previously it was displaying as a field of object and and was not using the custom component that we have created for the rgsf menu. So now it is using with uh proper additional properties, placeholder and items over here so, and it is also not affecting the containers and other environment reports and commands for the schema yeah.

E

So um that was a quick update for the pr and it's good to go after a review.

E

Yep uh any comments and like anything.

A

uh So does this take care of all the additional property skills, or do they remain remain as like the cuny new key thing from before.

E

uh No, uh it takes care uh for um the machine, this one uh additional properties thing so yep uh it should. This is the basic one that we are using for the whole rjsf schema component. That is overrendered in the pattern configurator, so it should take care of.

A

E

Rest of the two yeah.

A

Okay, so in this you have changed the properties in the schema right yeah. If you want to do it for other components, then changing.

E

A

Won't be difficult.

E

uh No, the schema the adjuster provides this functionality very precisely to do this yeah, because previously, when I was changing this um property directly in the ui, so it was changing all others uh field to this thing, so we were not able to add any numbers inside a field of array or anything else, so yep adding it to the schema is the only solution that we can use.

A

So that's what I'm saying so for other components. You would also have to add it in the schema right.

E

And yeah, eventually, we will be making uh each and every component as a separate custom component, so we can like have have set of properties and display it accordingly, so yeah it can be tackled in as a separate issue.

A

Anybody else any feedback or comments on this.

E

If not over to right, there.

A

All right, yep, thanks laura for sharing uh so moving on. We have the github action pr from harsh harsh. Would you like to share your screen.

G

uh Sure give me a second.

C

Can you see my screen.

E

C

uh Right so uh I raised a pr for uh you know: uh uh measuring pattern.

C

Apply we needed to uh have an action in order to test that, and maybe just you know, an action if someone wants to uh test their own independent action, so they can just spin this workflow, and this would uh give them an insight as to uh like in a dummy environment as to how it's uh going right so uh so here uh the thing is that this particular repository right uh this uh service mesh patterns, so this actually the the action in it actually refers to a different uh sorry. I'm just.

C

So this actually refers to uh I'll show it to you, so this actually refers to this uh particular. uh So this entire repository in itself is an action right. uh So this action.yaml is what this looks for and accordingly uh it's it's a little complex uh action, uh but otherwise it's simply that action calls main.js, main.js spins up a child process which runs dot, sh right, main dot. Sh does all your setup tasks then measure dot assets, it does all the uh setting up of kubernetes and uh setting up of measuring and all those tasks.

C

I'll show this one to you right. You can see uh it's it's uh manipulating, some like just minifying and flattening the cubesat config, and then installing and running, measuring, right and then finally, uh machinery ctl.sh. This has the commands that we finally run all right so at so.

C

This was something that was sort of partially done by a contributor before me, however, this entire flow was breaking and previous and uh in the in this particular repository I'll show this one to you guys as well in this particular uh repository, so uh in github for close pattern action. This is the one that we're concerned with this actually did not correspond to what is actually given over there. If I take you back there uh and I'll show you uh the action.cml, uh you see that this actually requires a provider.

C

Token, a platform input a pattern, url input, but when you see this, this actually is only providing your platform provider token. So this is sort of a previous version, which was uh the reason why there was a sort of a compatibility conflict, and this were close breaking so uh right now, uh if I can take you to the action itself, uh where is the action yeah? So here is the action so right now, this is in a working state.

C

However, as one extra step, what we're trying to do is we're trying to uh you know uh instead of sort of manipulating cube conflict. That was something that I showed it to you previously right. So what this one is doing it just takes. You know the existing or sort of a default cube config and it sort of minify flattens, and then uh this uh moves it back to uh the path, the required path.

C

So, instead of that, what we wanted to and what uh lee suggested that you take uh this cube config as an input itself right the way you're doing uh so I'll. If I can show it to you in actions just to apply a message, so branch.

C

Right so you can see uh these are the four things so there's a name of the pattern file relative. So there's the way you can uh do this as well.

C

So similarly, we wanted to uh give the person who's running this workflow an option of uh putting their own cube config in this, and we are not playing around with dummy configs ourselves. So uh this is- and also this is also a question uh or like uh to all of you. If somebody knows uh you know what, if there's uh some way of doing that, because, generally you know, these cube configs contain these sort of secrets and then path to some certificates etc, which are local to your own machine.

C

So, let's say in my own machine, so the path to certificate would be, let's say, user slash, my name, my username first.hello, then slash somewhere somewhere and certificate right. So I want to avoid adding those secrets and taking those secrets from the the vm that is running this itself and while also adding the cube, the user specified cube config. So if anyone has any ideas, so they can, uh you know, let me know uh yeah, so this was more or less it about this particular issue and uh what I had done right.

C

So if anyone has an answer to that question or if anyone needs a clarity or if I should repeat uh my question again, you can you can just let me know I'll be glad to do that.

C

This this is more or less set from my side.

A

Guys any feedback, any comments.

C

So if any one of you has been working around any cis or such so, if you are, you know somewhere in nec, because I I did not see this happening anywhere. But uh if there is, you know a particular workflow where you're supplying the cube config as a parameter, let's say path to the file or a url, or something like that.

C

So if, if you know something of that, so do let me know because that that will come in handy here.

C

uh Sure sure I'll I'll reach out to someone on slack, if you know, if anyone has the answer or you know, is interested to know about the problem or help me out, you could just hit me up anytime.

C

That shouldn't be a problem.

A

Would like to share your design updates.

C

Yeah sure uh just give me a sec and yep. Let me just connect away okay, we have everything: okay, cool, uh so, okay, that's not the screen.

C

So yeah uh I I worked on a couple of design improvements specifically related to the sidebar. uh um We are actually here to receive an act from uh lee. Regarding the same so once we receive the act from him, we will be able to implement the similar sidebar in our application, so in in our current uh version of the sidebar, the ui is not very good, and even if you take in consideration the user perspective along with user experience, uh it's it does not follow the right ui principles.

C

When uh let me just did that one go. uh So this is how it looks uh as of now and the proposed changes. uh Look something of this sort. So, okay, let me.

C

Yeah, so this is how okay sorry.

C

So this is how the proposed one looks like so, for example, if anyone hovers on it, it will uh kind of highlight, in the misery green, it's kind of like a overlay on the box, which is box or the specific text. So all of this will have that specific overlay and for uh the main headings it will have a small icon when it's covered, I'm not sure if you can notice.

C

So there's this small icon, which comes on the right side uh if I hover on the menu headings. So this is how the proposed changes to the sidebar looks like and uh yeah. It keeps in mind the ui ux principles as well as as we are using a darker background, the overlay or the highlight which comes on top should be a lighter one, and it should be of a color which contrasts to the background as well.

C

So if you consider the collapse of sidebar, it kind of looks like this on hover and again, uh the main header main menu headings have. uh uh This is a small arrows. I just noticed a small issue. It shouldn't show the gray background for that and then yeah all of these icons on hover. I guess it's already there. Currently it changes to a darker white, color and yeah okay. So this is how the notification center looks like okay and then we have all the notification categories or severe.

C

It is like error, so this is how the error part looks like, and this is for warning and this is for success and again, if I go back to error uh sorry all part, it will kind of look like this and if I go ahead and click this one all of them will go, and I can just collapse this uh back again so yeah. This is the short update from my end.

C

If you have any doubts or if you have any suggestions, I am open to them.

A

This looks great, so uh one thing, though: uh when we could you collapse, it.

C

A

Yeah so here like having the chevron beside the icon, don't you think that provides a lot of space.

C

uh Do you think uh here on the right side, uh next to the.

A

Not these, uh like the the small, like uh the small chapter on the side, all the icons like.

C

C

uh I don't think because one thing is uh if, if, uh if you are going to implement like a collapsed state, people should know that menu is, uh has sub options.

A

Yeah, but so right now what happens? Is the icon itself changes into a bottom chevron? uh Wouldn't that look better as well.

C

Yeah, I think uh bottom chevron like uh uh when we uh click on it. It will change it to. I didn't get you.

A

So right now in the navigator, if you see, if you hover on an item which is you know,.

F

C

A

uh The icon itself changes into a chevron.

C

Yeah yeah this small one: oh, no, no! Actually, uh okay, yeah! That's in the current implementation! Is it or.

A

Yeah yeah, like it's not beside it like the icon itself, becomes the chevron.

C

Okay, but uh another thing uh so I hadn't, uh I didn't get a chance to look at the current implementation of the ui. uh So another question: if this changes to chevron, how can you differentiate between this uh chevron and this chevron like? Is there any?

C

A

Know that happens that happens on hover, like only on hover.

C

Okay, yeah, I think uh chevron wouldn't be the right way to do it, because, if you, if it is in a collapse state and if someone uh sorry wants to share a screenshot of it, I think uh this would give a better perspective of the what sorry the main menu has has to offer. So, if you just consider it as a chevron, uh the chevron doesn't uh stand for anything other than it, so it just shows that it there's a drop down menu or it it has a sub menu.

C

Imo, uh I guess this would be a better option, but yeah there's one small issue regarding that uh background thing, which I will uh change it uh today itself.

A

Sounds good, okay, uh anyone else any feedback regarding this.

H

uh What happens when you click on the question mark I can like that is supposed to expand. It.

C

Yeah I I haven't uh currently implemented that one, so it kind of uh okay. I guess wait, wait. Let me just.

C

Is it in design language yeah? It is in design.

C

ah Where is the flow? Where can I start the flow.

F

C

So when you click on that one, it will kind of pop out like this and again it has all the hover states and all, and if you click it back, it will go back down.

C

H

Yeah- and I was asking because right now, the chevron you have used is, I think, is gonna cover the icon. It's a little bigger than the one we had currently.

C

uh This one right, uh the.

H

One showing.

C

This is it this one yeah, no, it it will kind of uh okay, I didn't think in that perspective, but if it uh overlays or if it comes behind it, probably we can add a state management for that uh something like. If you click on this, it will go to the outside part and if it, if it's not clicked or if it's not uh if the state is not set to open, it will just uh be here itself.

H

C

So yeah that's the definition.

A

Would you like to speak to your topic.

D

uh Hello, can you hear me.

F

D

Okay, so I'll just I'll share my screen.

A

Yep, please go ahead.

D

Yeah, can you see this game.

A

D

uh Okay, so this was the issue. uh The issue was that uh mystery ctl. Would uh it would just issue uh when you, ha when you give it the stop operation, it would just issue a delete instruction to the kubernetes client and it would exit and the problem with that is when you issue a restart operation, it uh issues that delete operation and then attempts to start back which then uh so it errors out, because the uh pods are still in the terminating state.

D

So what the goal was to uh wait until the delete actually happens on the names on the kubernetes side, so what I've done here is yeah. So this is what I've done here is just use this a similar logic that we use for uh starting uh like when the when you issue the system start instruction. uh It issues uh like it, creates the parts on the kubernetes side and then pulls the kubernetes client asking it if the pods are up. So this is what we do. This is what I've done when it stops.

D

When you stop the deployment, it pulls the kubernetes line, asking if the name space has been removed. So this is what it does so now, if I so right now, it's running I'm actually running on my system. So if I just say measure cpu.

D

The ctl would give the kubernetes client an instruction to delete the names. It will also wait until the namespace is deleted. So.

D

This takes a little while.

D

D

Also uh also, if so, it has a timeout of 300 seconds. So if it doesn't, if the new switch is not deleted within 300 seconds, it will just print a message saying uh that it's not stopped yet. You can check the status yeah as you can see there, it stopped it waited until it stopped and then now it's starting again.

D

So yeah, usually if without this patch, what would happen is uh after after it displays machinery has stopped, it would attempt to start it, and then it will fail. It would fail saying uh that one of the components is a terminating state and it cannot start.

D

Yeah so yeah, so, as I was saying it also, it has a timeout of 300 seconds if it doesn't stop within seconds it'll just print a message and uh it will tell you to check check the status manually.

D

So that's it from my side.

A

Guys any feedback.

A

So if the timeout occurs it uh like it stops the process right there right and you would have to run it again.

D

uh If the timeout occurs, it will actually continue in the same flow as it used to earlier. uh It will still go through with the start instruction so far. I've made it so that it still goes through with the like. The restart is made up of a stop and a start right. So if the timeout occurs in that restart process, while stopping if the timer occurs, it still goes on with the start process, but the start process fails, but uh but yeah the stop process logs a message saying that it didn't stop in time.

A

So we should probably handle that better right, like we don't want it to fail. So if we restart, if it hits the 300 second limit uh shouldn't, we probably stop it from uh starting again, because if it's going to fail on static, we should probably stop it right. There right.

D

Yeah, I think that's a good idea. We can stop it from going with the restart.

A

Yeah, so I I think that would be a better option if, if we reach the 300 second, we should stop the command like kill. The commander.

D

A

Anyone else any feedback or comments on this.

D

A

uh So we have a couple of call for volunteers, so, as some of you know, we have been working on a service mesh catalog.

A

uh So basically the catalog would consist of all service mesh patterns and filters, uh and I think even ashish brought it up and he was demoing that it would have the pattern files that you would be able to uh download later on.

A

Yeah, so this is how the catalog looks at the moment, so since this catalog would in include web assembly filters as well uh having two paragraphs, just you know explaining what patterns are, is not really a great design for the the top of the catalog. So there is a call for volunteers to design the top portion of the catalog uh there's an open issue on merchandise, for that and yeah, I think yeah. So there's an open need for designing the top part of the catalog.

A

Another open need is to design the patterns itself right now we only have three patterns and the rest are just that's why we have them under this coming soon overlay, because we don't have enough patterns to put on these cards. So there's an open need for someone to understand how these patterns work and to design them. So if any of you are interested in doing that interested in working on sigma or ux, as you can see, it's written here uh see like leaky code. If interested so yeah.

F

A

Since we have five minutes left I'll talk about the refactoring of measuring ui, yet again, so we are currently in between of refactoring measure the ui for the v 0.70 road map. It's a complete overall of the machinery ui in which we are incorporating redux as state management.

A

As you can see here, we have a branch in which this uh effort is going on. So it's called the missionary ui restructuring branch uh here, if you go to the contributing md you'd, be able to see the contributing guide on how to set up your, uh how to set up your environment to be able to run this uh server and be able to contribute to the uh contributed this branch there's the proposal on state management machinery uh state management by using redux. You guys, can go through this.

A

If you guys have questions about why we are using redux instead of any other tool- and there is also an implementation tracker. So this basic this google sheet, basically has all the reusable components written down over here. uh We have tried to cover a lot most of the components, but there is a high chance that we have, mr couple. So if you find something that is, that is something that is a reusable component and is not listed down here.

A

You guys could uh write that component down, put your name against it and start working on them and what else yeah, so that's about it from the restructuring. So if anybody else doesn't have anything to add, I guess we can call it a day today.

A

Anybody if anyone has any questions they want to ask you guys, can go ahead.

A

Okay, three second limit: nobody has any questions, uh so thank you guys for joining I'll, see you guys next week at the same time, and on and on friday as well, we have a community called coming up. So please join that one as well. So thank you guys for joining bye, bye good night good morning, bye.

H