Kyma Core Special Interest Group Community, 11 Sep 2019

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Kyma Core SIG community meeting 20190911

Description

1:23 - AddOns configuration on a namespace level
9:53 - Kyma 1.5 got released and what it brought
13:58 - Upcoming Kyma 1.6 scope
15:53 - Documentation rendering component
24:01 - Kyma CLI supporting provisioning of Kubernetes clusters and Gardener Terraform provisioner

Meeting notes: https://docs.google.com/document/d/1vWleTon7sJIk0teee4SoVezS4mR3K8TlkvXkgLJwRD8/edit#heading=h.5daajeb8h8vr

A

Hello, welcome to our Korsak meeting I Mukesh and today, I'm gonna, be your moderator and mater whoosh. That has no better place to sit so he's sitting on the streets in war, so he will be the guy. That's helping news taking notes before we start remember about our rule. Add your name to the list of the attendees. So we know that you've been here and we can share the info to others.

A

Today's agenda is pretty full of different things and there's a small change in the agenda on request from by Mike sure stuff. So we usually start the meeting with asking people that we see here for the first time how they're doing, for example, but we don't we're not gonna, do it because shake has similar meeting, and so he will actually be the first today to present during the meeting and then I will I'm gonna go back to the agenda and handle the agenda and the rest of the meeting with you.

A

So please don't feel disappointed and critic. Are you ready? Yes,.

B

A

Are something I'm stop sharing and please do.

B

Can you see my screen I.

A

Can see yeah no I can see it. Okay,.

B

All right so hello, my name is chick and I want to show you today a new possibility. The possibility that this that emerged in kima, in in version 1.5 and in the UI you are able to manage the content service catalog and in a given namespace.

B

So for the purpose of this demo, I have I have here a Kimmel runtime running on jegi cluster and I have two namespaces here: the default one and my own use Redis here.

B

So if I go to the any of it and open the Service, Catalog I can see two service classes here and if I go to the other, one I can see exactly the same service classes in the other one, and this is because in currently on, my runtime I have a cluster wide set up for the Service Catalog I have a cluster white add on definition that defines a remote repository of atoms and those atoms are then transformed by helm, broker and are available to the Service.

B

Catalog and Service Catalog creates service classes for each namespace there, for each namespace has exactly the same service classes.

B

Now, with you probably as an cloud administrator, you would probably agree that it's not always wise to expose every resource everywhere in every namespace, especially if there is some cost related to consuming some services.

B

So in order to provide a better control over which service classes are available in which namespace and there is a new flow available that allows to create an admin configuration in a given namespace and then in this flow helm broker would identify that there is a new configuration for a given namespace and will fetch all the information about the Allens and transform that into service classes in a given namespace. So, let's, let's see, how does it work in action.

B

Ok, so, instead of simply adding more advanced configurations to the class cluster white, add-ons I would simply go inside my user IDs here namespace and choose add-ons navigation option here on the left. This is the brand new view, but that is now available in kima 1.5, and this allows me to create a new configuration just for my just for my namespace and for this demo. I will just take a testing bundle from a atom repository, came an atom repository and copy this link.

B

What's inside it just defines to two service classes, testing a done and ready service.

B

And so I'll use this URL and configure it here as a namespace names, paste atoms, config, atom configuration yeah. So what happens now is that I have a configuration just in this namespace and additionally, a new broker is available, just in my name space. That is this publishing those two service classes, definitions for the Service, Catalog and I- would expect now, if I go to the catalog in my namespace, that I would have some additional classes.

B

Yes, exactly those which are defined by the which I added so I can now use in memory database ratings just in in this namespace and those service classes are not available in the other namespace. It's still tested default, set off of service classes, as defined by the cluster white configuration.

B

So that was pretty much it from my side. Do you have any questions? There's a documentation about this flow I will send it in the chat after I can read more technical stuff about how this all this magic is happening.

A

My question from my side curious so did I say it writes that, like in the diagram, the diamond broker creates a brand new system instance of a service broker inside the night space on a first addon resource creation right. So normally, when you go to a brand new in the namespace, you there's no rock.

B

Surgery in there, okay, exactly.

C

Basically, there is only one instance of service service broker the hand broker seats, undercook kima system namespace, and we only adding new endpoint for serving additional service classes for given nice place. But there is only one application in your cluster okay.

A

So you're adding just creating a service broker resource exactly.

C

Okay, new endpoint cos it.

A

Okay, any other questions to you.

A

B

Maybe I will just mention that we plan to expose some new features on that view in the next release to provide better visibility about statuses of the add-ons and also the option to refetch the configuration, so that would be presented, probably in the next six core meeting as well.

A

Okay, awesome thank you again for stuff and yeah run run right, enjoy the.

B

Rest of the anything okay.

A

Okay, I assume you can see the agenda back again.

A

Okay, so next item of the agenda, questions and feedback. So we, if there's someone in the meeting that would like to ask a question before waiting to the end of the meeting, because we also have something like just off and you would like to leave earlier any questions.

A

No, some I see that Ian you joined, then it's awesome, some so enjoy the rest of the meeting. So the the next items of the on the agenda will be quick intro to the release.

A

We did on Monday, also a bit I'm going to describe a bit a code for the next release that we already you know we have this cope, so we can share something about it, so something that will be released in next month and then I will do a boring presentation about documentation, components that we did and, at the end, we're going to have Borja with topics around terraform and provisioning of Kimura and kubernetes clusters.

A

So first is the of course the release. So let me simply go to the release notes.

A

Before I jump into the details of the release, notes information for you that on Monday we did release on Tuesday. We also already made sure that kima 1:5 is available on Google Marketplace. So if you have a tcp account, you can yeah much quicker provision in map to play with it through the Google Marketplace.

A

Then, if it comes to the scope of 1:5, what we delivered the most important things I think to notice is that we first of all make sure that we are compatible with the latest Barnabas 115, but just bear in mind that, because of the invitation on gke and aks that do not support such a high version of Burnett is yet. We only have a automated testing on cornet is 1:13.

A

Then, if it comes to the important topics, we're super advanced with in implementation of the of our API gateway to extend it with more functionality, especially oauth2 support.

A

So it's already in a scope for 1:6 to complete the implementation. So it's gonna be much easier and much much nicer or more more options. Simply you're gonna have, when exposing applications to the outside world console no need to talk about it, because you could see now a presentation from Christophe, so in exodus level, Adams configuration, then we in documentation I think important is that we have some nice new tutorials for monitoring in a service management.

A

That's super interesting as especially that today, peeled, who is a product owner for service management in public in key mark, is doing a presentation actually I think in half an hour on handsomest in Amsterdam, and he will probably announce what we are now simply announcing on the meeting.

A

So the the fact that we now officially maintain hand broke outside keema repository, um so we will now officially always tell people that you can use hen broker in under scenarios, not only in keema, because we can see that it could live in the world of service catalog without kima, and we could try and attract more people and to cut contributions to the project. So that's super important to notice, then with for people that are super interested with the key native, how we're going with the implementation, with this release in keema?

A

Hidden by the future flag is a new experimental function, controller that will be a replacement for our cublas support in for surveillance in in c'mon. So if you would like to see how it works, how it's consuming a key native built and key native serving yeah, please give it a try, enable it and play with the in with the controller and the function of our custom reserves. And last but not least, we updated is Joe so for people that are concerned about security.

A

Please keep in mind that we have now a much newer version of this Joe, which is solving many issues, security issues and actually the same goes with the with our support for kubernetes, because the there are some major security issues with kubernetes below version 113. So now you can easily use among the latest versions.

A

So that's if it comes to 1/5 what is done and you can play with those features already important, is also to know. What's in 1 6 again in one month, first item sorter form based provisioning of kubernetes clusters.

A

We started a incubating project called hydro form. As that you can yeah. You have a link in the meeting notes.

A

You can check out what it is, but I think it's really interesting to see what direction we're taking with provisioning that we want to basically team our CLI to support also provisioning of Governments casters using terraform here and then, if it comes to kompis project, another incubating project we have in you know, working my incubator organization, we yeah plan to have the provision provisioning of kheema handled in the encompass, as well with one six release and two more things lock UI, if you're a long cookie, my user, you already know that long Qi in keema is not the best one we're getting a lot of feedback that the usability.

A

It's not what developers expect and there are there's a work plan to improve the UI and last I already mentioned when I was talking about 1/5 release in 1/6. We want to complete the implementation of the API great way to have a of to support using the projects from Ari called Hydra.

A

Thing any questions before I jump into next item about 1/5 or 1/6.

A

You you, okay, no questions so more fun. Now, cuz I'm gonna do a presentation of my beloved documentation component, so I prepared a really boring presentation and to do it, but I think it's just this way of explaining what the condition component is and why it's important for you, because I think I own you first of all a context why we created this component, because intentions were different for us and we have a different result and I don't want to show you in action how the component works, because that's also not important.

A

It's super easy react component. You should rather just notice that we, we simply have it open sourced, ready to use in any UI, and so for you important, is to know what it actually is. So first slide.

A

If you you're new to keema or even existing user I, think the basic understanding that you should have is that in in keema, because we yeah want to provide this extensibility of monoliths and you can imagine- and they are sometimes implemented with different technologies, and then they expose api's with different specific use in different specifications and standards in keema. One of the rules we have is that we yeah we value specifications and we don't want to reinvent the wheel.

A

So you in this simple pipe picture you can see, on the left hand, side a group of monoliths that exposed open, API or data or asking API specs and in the key McCluster, can see something done with the markdown specification and also if you could see on first of presentation how he could show a service catalog, all those services like Asian broker, etc. That's all powered by open service broker API specification.

A

So let's let scheme up right so vision of Hema is to support standards, don't reinvent. So as we result what we have, for example here, you can see a a view of the documentation section in the console of the key MIUI, so you can see that we, for example, provide a baked in an documentation for kima, so we don't have to search for information about kima, ending on the internet. You get the docs for latest version of your cluster inside the console.

A

We also have this contextual help. So, for example, when you could see on a presentation from Chris you, you could see that he's having some classes in the Service Catalog and, for example, there was this Asian broker, so then user. When he's more interested in with the service, and he enters the service, you can see additional documentation that is showing the context of the given service in this screenshot.

A

You just see markdown files rendered I, say I see documentation, but you can imagine here that if it's an a service that is exposing API or is sending events into the kheema cluster, then you would see a tab with API console with even documentation rendered also in the context, and then, if you know our kima project IO, we also render a lot of markdown there and it's a little bit different differently styled.

A

The navigation is a bit different and the scope of functionality for rendering markdown is smaller than the one we have in the catalog and view in the console.

A

So we can imagine that when we started kima- and we started thinking about supporting all those specifications, then what we did first, we of course did the research we found a perfect match for us for open API, which was swagger UI for markdown. We found react, markdown, which was perfect and for us in API there was nothing 1d on the market, so we basically wrote a new component from scratch and invited into it to as an API organization and for all data the same.

A

We had to write a component from scratch and keep it inside, but then, if you could see, we have a console view. We have website repository with team, a project I/o, so as a result of reusing those components in the views we had a lot of code, duplication, yeah. We had to implement things.

A

You need a placated way in two different repos, so we basically were. We were wasting a lot of time, so we decided to create this documentation component to yeah, get rid of all this code, application to make maintenance easier, but as a result, what we notice is that okay, we we created. Basically, a generic component that can be used by anyone in any context not related to him at all, just for rendering markdown or api's api specifications.

A

So here I'm, just gonna share you two links, first of all to the documentation component repository if you're interested with the subject of rendering the different specifications go there and see how the architecture looks like because it's really flexible and gives you a lot of accessibility options with our plug-in system. So then, please do it and give us a nice start to the project because we like them and then in the presentation, you can see two links.

A

I'm gonna share the link to the presentation in the agenda and I think the links I also provided already in the agenda. So you can see a sound output from the sandbox project that I was creating some time ago and turned it up with matches from my team today, because we released zero to zero release today with default styling. So we can see that basically the whole component, that's it.

A

In this example, you can see that I'm having some mock data, so I have some mock as an API, a mock, markdown, OData and open API, and this part here below this line is everything handled by the component of course. Here, for a sake of these sandbox I created a simple navigation, so you can see as an API rendered OData rendered open, API and markdown, and, as I said today, we released zero to zero, which means that we also include default.

A

Css is based on ASAP fundamentals, so you can imagine how you can see how easy it is, for example, to get rid of default styling. So here in the markdown, you can see some nice panels the copy button for the code snippets, but it's and thanks to injecting our custom styles.

A

If you just don't, don't want to use them. You want to use yours, then that the component will just give you a basic stuff that you can style on your own.

A

So that would be it.

A

Do you have any questions.

A

A

But you're still there right.

A

A

I guess you are, are you damn? I was actually scared, so thank you and now I will be quiet and let more huh to show his awesome stuff.

D

Okay, thanks for cash, let me share my screen. First of all, so I'm going to show a couple of things, the first one being the integration of the kima CLI with terraform, for that we built a concept based on Google cloud platform, which allows us to deploy kubernetes cluster stand, specify them via the CLI and then the implementation uses terraform. So you create the cluster and manage it on gke on the kubernetes engine.

D

This is still under heavy development, so I'm just going to show the main feature of this, but it will probably change a bit from now to when it's done. So, let's take a look. Basically, we have the Hemus realized, so I'm going to go to the console Rakeem, our CLI is basically the kheema command, and for that we have a provision command that allows us to to create clusters in different platforms. If we go to Kemah provision, we see.

D

Okay, we have can send that cluster into mini cubes or local development or create a cluster in GCP. This is the the concept which is created, so let me check what UCP says.

D

Here we have the options we can pass to the GCP list, so most of them I'm not going to give because they are already default values such as the kubernetes version and machine type region, but they are configurable. The important properties right now first is the credentials file. This should be a adjacent key from cloud platform for a service account with enough rights to create rosters and, secondly, the project so in which Google project we want to create the cluster.

D

So, let's check I have here my project, which is the kheema project we can see it in here and in the URL, and thus I'm gonna run the command as kheema provision DCP, so I need, first of all, my credentials.

D

Which conveniently is in my local directory, but it could be anywhere, and the second thing I need is the project name, which is as we check in the URL Hema miners project. So that's our project.

D

So now, when I run, this I should see a cluster, it's starting the provisioning. It will stay like this until the cruellest arisen and if we go back to the GCP and refresh, we should in short, see a cluster with the name terraform being created there. We have it. This is the cluster it's created with the region. We gave the version. We gave the defaults, of course, and that's it once this cluster is finished.

D

We can just use it. That's it simple as that we aim to have this for more providers and the second one we have is still not integrated in there and the CLI itself, but we do have the terraform integration for it, which is the gardener provisioning. So for that, let me switch to the gardener dashboard. This is my dashboard and for that we have a terraform provider that will manage their creation, destruction and status of the different resources.

D

So I have a project or an called Borja, so my project, as you can see, I, have no clusters on it right now. Here I can manage my keys. I have a key for Google cloud platform, which is this one, and these are my service accounts for a gardener. I have bottie, which is my admin, but and allows me to manage clusters on gardener. These things are what I need to to to use the coordinate provider. So let's take a quick look at that I have here my.

D

Terraform for the gardener provider, the first part is the provider itself, so my credentials like in which profile do I want to create or manage clusters which secret I want to use to access the underlying platform, and this cube path is the gardener cube config. It corresponds to the configuration of the service, but the body account, which is the admin that will in my name, perform all the operations and the other resource we have is a gardener, GCP cluster chute cluster usual specification name of the cluster region, some subnetting for gardener.

D

This is something we will not go in detail, kubernetes version, type of machine and so on, and so on. So basically to run this. The first thing I need to do is to run terraform in it, so it can basically generate the the trough form files and the state files and check that I have the plugins necessary to run this. So once terraform in it is run, we can see that everything is fine cool. Let's check the plan and plan is telling us everything. It did a dry run a test run, and it's telling us.

D

Okay, I am going to create these resource and that is going to be what we're going to do when we do to reform apply, which does basically the same thing, but it carries out the operation so once again it it gives us a summary and asks for our confirmation, I say yes, cluster created, let's go back to Gardner and see if it's true- and here we are- we have a governess cluster in the region.

D

We specified with the name were specified kubernetes version we specified once this creation is finished, we can ask for the access and we can start working with it. So that's it. That's all for the different provisioning options for the kima CLI, thanks for watching and I, give back to you we'll crash I guess.

D

Unless there are some questions.

A

Thank you for her any questions from the audience.

D

One two three actually.

A

One from my side, but because you showed now using Terra from directly but I, assume that there is going to be something like chemo provision, Gardner PCP or something like this correct.

D

So that's why I showed both things, because the first one was the integration of terraform into the CLI and the second one is a provider, a Terra firme provider for Gardner, the provider. I used it with the turf from CLI, because it's easier to show and because we don't have the integration yet into the CLI. But we will. It will be exactly the same as the first one. I showed just for Gardner.

A

Thank you. So there are no questions. Let's just summarize the meeting final one minute to ask questions.

A

You sure it's better to ask them now than on slack.

A

E

I I was kind of in it when's the Ling cubes are the things kicking in so.

A

The mini cube is supported as far as I know right. It's.

E

Mini cube and that's already used in terraform right.

D

So the the mini cube provisioning is not using terraform as of now, because it was the first thing we support it actually and it's just creating mini cube locally by using the actual mini cube, CLI yeah.

E

Okay, cool, so that's one of things on the to do. I guess is it migrated over to terraform just to give you some consistency? Yes,.

D

So, in the end, the the the whole cluster creation layer should be completely abstracted for us and and and be completely based on the reform for any provider. It should not matter cool.

D

That's it for me, yep thanks. Thank.

A

You so thanks a lot. Let me summarize again, thank you so much for joining I hope you enjoyed it next meeting in two weeks. Remember that you can always add your items to the agenda upfront, especially if you want to have some a expert in some specific fields that can come to the meeting and answer your detailed questions. We can always organize it like get some dirt from a cave or something for you. We have many of those. So thanks again, thanks to the presenters, see you and hear you next time.

C

Yeah my goodbyes.