Kyma Community Meetup, 15 Jul 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Kyma Community Meetup 15.07.2020

Description

This virtual event is designed for end users, a space for our community to meet, get to know each other, and learn more about Kyma. We plan to have at least one demo in every meeting, with allocated time to discuss this in small groups.
Agenda:
🌱 Pearl Group would be happy to share our journey and experience with the Kyma product. They have built an integration template to connect and exchange Open API and Async API with our Loyalty product and Kyma (Pavels Koliskins, Pauls Zunda) - 1:40
🌱 Local installation of Kyma on k3s cluster (Piotr Bochynski) - 37:52 https://github.com/kyma-incubator/local-kyma-k3d

Get the meeting agenda to find out what topics are planned for the meeting.
https://docs.google.com/document/d/1vWleTon7sJIk0teee4SoVezS4mR3K8TlkvXkgLJwRD8

Find out more about our meetup at : https://kyma-project.io/community/collaboration/#kyma-community-meetup-kyma-community-meetup

A

Hello, hello, everyone and welcome to our keema community meetup. We have a special meetup. Today we have a couple of people coming from outside of the team presenting their projects, um pavel paul, palvel collins.

A

I practice and I didn't manage. Okay, pavel collins, kolinsky college.

A

Okay from pearl technologies, uh they are going to show us their projects and how they are using keema uh for for their team and for their product. That is going to launch very soon right.

B

uh Yeah, so it's a ongoing process. Now, yes,.

A

Yeah exactly and then we have, if the time of allows, we will have a a demo, a very interesting demo uh from pb computer regarding the local installation of keema using k3s, which is something that we are working lately a lot in the optimization of the installation process, and this could be an interesting way to to to install it on your on a local machine or on a lightweight version of it. So, let's, uh let's not keep everyone waiting. uh This is the agenda and I give now pavel the demo the screen sharing to you.

A

A

Perfect so let's see what uh what we have.

B

Thank you. uh Do you see the screen.

A

B

All the presents you see the screen, so yeah, hello, everyone, it's a big pleasure and honor being here uh yeah. uh So basically my name is pavels, I'm coming from pearl group and I work as a sap, uhana solution, architect and today I also have a colleague with me paul yeah,.

C

I'm pulse yeah, uh I'm not more a technical guy regarding the sap, commerce integrations and uh keema the development side for it.

B

uh So essentially, we prepared a few slides, so we just wanted to give some baseline about our company about our product and then dive more deep. uh How we actually use schema and uh yeah share it. Actually our experience and also show the demo. So hopefully the demo uh will work fine, uh so yeah. Basically, uh we we have pearl, so we are sap, uh platinum partner and uh in addition, we have also partnership with abs and asia.

B

We actually have six offices in five countries, so we have two offices in norway, one office in sweden, office in latvia, ukraine and singapore. So we are around 190 employees at the moment and uh yeah. Our core competence essentially is s4hana projects, uh it's c4hana, so it's a commerce, marketing sales and service and in addition, we also do like analytics and integration projects and and of course, also basis operations on top of that and when it comes to our strategy.

B

So of course uh we always like aim to do and take total responsibility uh towards our customers, and also we are striving having a long-term uh partnership with our customers.

B

So yeah being said that so again, uh our like key focus. What we work on is s4hana, so we started uh migration project uh a long time ago and just asked for hana release so being one of the first ones actually in europe. Doing such such kind of projects and yeah c4hana is also a big big thing so, especially especially the commerce and marketing cloud. This is where we focus on yes, but uh of course, yeah sap is one thing, but um actually what we did uh recently.

B

We partnered up with another norwegian company uh called sunsen, and we started to build uh additional products, uh not on top of sap, but uh additional products which could extend the sap portfolio essentially and and we branded under uh cxpi uh suite. Basically, so we are actually working in four directions, so we have our own.

B

Like recommendation engine, we have the add-on for apache solar, which brings like semantic search, uh personalization and enhanced analytics, and then we have the loyalty product and we also working with the data and analytics uh area, basically and I'll a bit stop uh by loyalty product, because essentially this is a product we use to integrate with schema. So just yet briefly uh tell what it is.

B

So, essentially, if your enterprise is uh looking into implementing some loyalty program, so that would be the product we would, we would suggest uh it actually has like a flexible rule engine where you can define different, like loyalty rules, how much loyalty points by buying this product within this shop within this store things like that, so of course, you can configure uh different uh account level and tier levels and yeah, of course, of course, as being sap partner, we want to integrate all the products with sap, so this is this is what we also did here and yeah.

B

This is just a short insight. How how it visually looks so, if you're familiar with sap, it also uses like sap ui5. So it's also pretty pretty similar to all sap like look and feel products, essentially yes, and maybe just uh yeah, giving a short background, how we ended up actually being here, um essentially um uh yeah, as mentioned, we.

B

We have, of course, our um our loyalty product and um basically uh in in the beginning uh being as sap partners, we wanted to uh connect it, of course, to main sap products such as sap commerce. So we started by building an add-on extensions on sap commerce to like replicate all master data and transaction data to the loyalty product, and when this was done, uh we thought okay. So um what else do we need?

B

We need, of course, our customers to be able to uh like enroll exit from loyalty programs see their point, balance, see their transaction and, of course, the perfect use case, for that would be to build uh like an add-on spartacus so which is a sap commerce storefront right.

B

So this is what we did as a next step and we were very close on uh starting and building direct integration with marketing, but we realized that there is a keema product essentially and maybe, instead of building uh direct one-to-one connections between our product and all sap products, maybe instead for us would be uh better to actually uh connect our loyalty product with schema and then use the power of like lambdas of chemo runtime, to extend, extend actually all these business processes and and connect uh uh other sap products from keema side.

B

So this is actually um how we get there. So we actually stopped the initiative of doing like direct integration uh at the moment and instead we uh build the keema connector on our loyalty pro product where we actually implemented the system airing uh to uh keema and we also exchange event and api metadata. Basically, I forgot to mention, if you have any questions, just just. Please stop stop us at any time, and just just raise your questions so that that's okay to do that.

B

So what we actually did so yeah, first of all, uh our integration now supports keema 1.13 version and uh it's kind of pretty similar as as it's done on stp commerce, like one click integration, so we call it in the same way, essentially uh yeah. Just you just copy, uh like uh the connection string, the token and to our loyalty product and then uh the system does it like all automatically, and then we exchange a synchronous api. So from a synchronous api today we support actually four types of events.

B

uh So that's a new loyalty customer registration. So when customer registers in the lower the system we get an event to keema, uh it's a customer unregistration from the loyalty program, it's any type of membership activity uh which which might happen, and that's also the tier level, upgrade event. So when customer changes the event, for example, from silver to uh to gold, then the event will be sent out to keema, basically and from open api perspective.

B

Here we also support several open api event, apis essentially uh possibility to enroll or unregister customer from the loyalty program. It's of course getting all all customer details, such as basic information. uh The point history uh point balance and uh also a possibility to publish uh essentially any membership activity like which could give extra point or essentially redeem extra points. Basically and maybe pulse you can do some more technical uh view. uh What was done.

C

Yes, so before I dig into these uh code snippets or graphql snippets, uh I won't tell that we- we mostly are coming from this uh java spring based background, so meaning that we didn't have much of experience regarding these kubernetes and all that so having this successfully implemented means that at least encourages other actually from any background, try and try and set up this schema with their systems.

C

So what that's? What we did? We had the spring boot based application.

C

Which is this say, xfi solution based on and uh we just uh yeah. This is just going through. This uh tutorials are step given uh in the keema documentation, and although, if you see the step two and step four uh in the keyword, documentation is written using the open, ssl toolkit, we basically replicated that idea in the java and that's how it works in the first step. We we just get this signing request for the certificates and we do it in the java.

C

We create this uh certificate signing request and we just push it or ask keema for them to sign it, and we receive these certificates in step. Number four, which we then later use for uh oncoming communications, and what we also on top of that have is that we have this job which uh renews these certificates, because uh they do uh outdate after uh some time uh and that's a step one on the step. Two of our one click integration so to say, is that we add a package.

C

uh I know that you can do it in one mutation, uh add api and event definitions in one, but we opted for these separate steps. And here we have. We use loyalty package by default and we also give these username and password. As you see in a default default instance, authentication which is generated on fly and for each uh package there will be different authentication.

C

uh What we did want to utilize was this instance alt request into schema. We know- or I think we understood the idea behind it, but unfortunately uh I didn't see any examples of it. So for now it's how it is, but for future it would be great to have this, for instance, use and then next slide yeah as step. Three is the uh api definitions and mn definitions, and uh this is uh and yeah. This is, we just uh add our apis and event to this loyalty package.

C

uh So this is. uh We are also not sure, if that's the best case scenario to put api and event definitions in the same package, so uh maybe by separating them there's some, maybe better uh or logical improvements on that. So that's uh that's! That's uh uh that worked yeah, that's our basic integration with the keema and uh and so as I said so, if it's possible to do from java based background without prior kubernetes knowledge. I I bet that anyone from any background could do that.

B

Yes and uh yeah, just just maybe um uh to give also one extra slide on on what we did on top uh within this product. So, as mentioned, we essentially connected in the beginning, uh sap, commerce, uh cloud and spartacle storefront, so the direct interface integrations. Actually, you cover the master data customer product category replication, uh it's possible to register and register from the loyalty programs and purchase activities, view membership, details and loyalty points and also actually redeem the loyalty points to vouchers so yeah.

B

This was like in short, on what we did in a theoretical level, so uh we were actually prepared also one demonstration use case. So if you don't mind, we could proceed with that unless there are any any questions so far for anyone.

A

B

A

Put the questions at the end and then so we can, you know, give you enough time to to cover the the demo.

B

So yeah, as mentioned, uh we didn't uh really connect marketing cloud directly, uh so we went in the direction where we plugged everything to keema and but still we see a lot of use cases where we will need to connect our loyalty product with marketing cloud, so essentially as a proof for ourselves that this was a wise way to go. So we actually have. I mean I mean quite probably a good use case, as we think at least uh so.

B

Essentially, the use case we are covering here is the customer who registers in the loyalty program. uh We actually want to give a special incentive to that customer. So, after customer enrolls in the loyalty program, uh we want to give additional discounts on the commerce in terms of the coupon and then we also want uh to send this information like a welcome letter and then send that coupon details actually as an email to the customer. Just saying, thank you welcome and here's your coupon code. You could redeem um so how how we actually achieved.

B

That is that, uh of course, since we have our loyalty, product and hema connected, we are getting some events. We are uh actually uh plugging into the membership registration event and then we are creating our lambda function in the keema. uh So when we get this event, we are basically uh reading uh the customer details from the local system. So we need to get the email and then and other customer details and of course, when we have it the first thing we will do.

B

uh We will assign the coupon to that customer account in the commerce cloud. So commerce cloud has this personalized coupon capability, so we have the coupon which will be linked to that account and customer could see it in my pages. Basically, and as a next step, we will need to submit an interaction, uh loyalty program, registration, interaction to marketing cloud, uh which in turn has a predefined campaign, a trick, your base campaign, which will already send a personalized email to the customer.

B

So in this case yeah we actually will achieve this customization without touching any of the systems actually directly. So we will uh do it uh side by side, the extensibility way so to say perfect. So I'll try to go to demo so yeah, hopefully, everything works. Fine, we did kind of a drive run, but you always know how it how it is right when, when the demo starts so uh yeah, I'm actually at my um I'm at my uh our cx.

B

If I exit by suite I'm on this particular storefront, I guess some of you might might have seen it before so. The first thing I'll do I'll, essentially login as a customer, so I have have a customer account already pre-created.

B

And yeah after logging in essentially uh I I will see different my pages. I can first go in the coupons just to see that they don't actually have any coupons assigned to my accounts and then I will go to our like loyalty add-on, which we enabled- and this suggests me to enroll into our cx5 parallel norway program. So as soon as I trigger the enrollment process, what will happen? The essential loyalty system will receive a loyalty registration event through direct integration and then uh well.

B

The system will trigger uh event to keema and then all the lambda functions will process. So what I will do yeah I will. I will open also my ssh, which is connected to the keema cluster and I'm now actually looking for uh my lambda function to see what's inside, of that, so I'll just just go to that, and then I'm enrolling, yeah, okay, I'm enrolling and somehow I didn't get actually an event.

A

A

Can I ask what version of kim are you guys using now.

B

This is a 113. okay,.

A

Actually, there's a very nice logging view in inside of keema in the lambda directly. If you you guys, have it so you don't have to use the terminal. You can see the logs directly there.

C

Yeah but we are using.

B

C

What we are using, we are using this uh amazon uh local uh setup: oh yeah, yeah, okay, okay, I got it so just to be very limited on resources. We basically cut off uh all the non-monetary things, but we had it for a while, but we were deploying a lot of namespaces lambdas and it consumed available resources. So we just.

A

Well, this goes very well hand-in-hand with the pb's demonstration, so we are trying to uh make it slimmer for for running on laptops. So that's why k3s will come in and uh also uh you will be able to in the future, just choose couple of features that you only need for a local development. So then, hopefully you guys don't have to do this hex too.

A

Okay cool. So you see.

B

The event yeah actually now it worked. I don't know why the first one didn't work, but, as you can see now, we received this essentially the event and uh yeah essential one. The function has received that I'm enrolled. I will go to coupons and you know allegedly yeah, I'm now seeing actually my coupon account available here and uh yeah as the next step. I'll just also go to the our marketing cloud system.

B

uh I will try to see the you know the contacts uh and when I go to the context, I will find my contact with which I'm logged in and allegedly yeah. We should see a loyalty registry program, registration event, uh which is also here. So what will happen next step uh marketing system will process that event and we'll send out email. It might take a few moments. Maybe before we we are waiting for this. We can go directly into keema and we can show actually our lambda function and how it looks.

B

Yeah, so we have our cxfi uh namespace available and uh what we did. We actually plugged the in the service catalog to application, so we plugged our cxfi application. This is our custom connector, which we did and then we plugged the standard, sap commerce application, which we also run like side by side and under the lambda function. Basically, we have our.

B

And we need to allocate more resources. Apparently we are running our cxfi uh loyalty, lambda function, uh which essentially, uh what it does. It essentially receives this event. So this is the log we we actually saw and and then what we are doing here as mentioned, we are getting that customer id which is send out with that lambda function. uh We are calling and a loyalty endpoint to get the customer details. So essentially we are interested in the email, so we're getting the email of the customer account back and yeah.

B

Then we are calling some old data services to assign uh the coupon uh to the uh in in the ccv2 in in the class to the customer commerce cloud uh to that uh email account essentially. So this is what actually was triggered and as a next step, we have two requests actually to marketing system. So one we need to get some xsrf token and some cookie data, which is a must to pass in the post request.

B

So we're just doing one get request to get the token and the cookie details and then we're just essentially populating this interaction data. So we are feeding that customer email address. We are sending lawyer registration event and uh yeah. Basically, that's it. Then we are submitting that and along the function essentially is over.

A

So but this part at the end, because you guys didn't make the integration with marketing cloud right. The whole.

B

A

Didn't have it manually.

B

So one option would be, of course, that we would uh like directly plug cx file loyalty to marketing cloud. But but again, if you look at from our side, we also start to think now, uh then we might also want to plug it to sales or service cloud.

B

In addition- and maybe then it's better like you know to have this connector to keema and then you could easily plug more systems, and maybe customers have their own in-house system, which again brings this kind of agility for us, so I think the email was actually sent, so I will just show once again refresh my contact.

B

I think it's it's caching up. I need to go to other one.

B

Yeah, as you can see, we got second interaction which is email, so it means that the campaign was actually triggered and if I go to my essentially gmail account, I see I got some welcome to cxfi loyalty email, which again is personalized email, which includes my name surname and also includes the loyalty code which I could then consume in in the commerce cloud.

B

So if I'm basically going back to the commerce cloud, I'm starting buying some stuff, for example, I essentially add item to cart and if I go to the checkout yeah, sorry not to check out but to cart. uh It essentially essentially should be able to consume that loyalty, coupon code, basically here and finalize the order. So this is what we also see under under the coupons here, so the coupon has been assigned basically.

A

Cool full integration looks nice.

A

Full circle so uh looks good. So should we start with the questions or you guys, you guys are ready with the demo.

B

Yeah, I think so it's just uh just wanted- maybe to show this one part so which I showed on the slides already so yeah. This is our package. So these are these four events uh which we are supporting, uh payload, which we have here. uh So I think we we received customer register events so yeah, it's this one. Basically, and then we have also yeah this open apis, which essentially yeah allow basically uh uh to see loyalty, point to see point balance to get customer retail and and register membership activities.

B

Yes, I think that's more or less right old pulse from our demo, at least what we wanted to show so yeah. We are open for questions and hopefully we'll be able to answer them.

A

B

Already have a positive.

A

Chat people, congratulations, sending congrats and, and looking good. I guess mostly. uh You already talked with with some of our colleagues uh with marco for sure uh we do have one question already from john. uh Basically, he says: congratulations looks good. uh He asked about the one-click integration with keema if it's java based and if it can be easily used by other java applications.

A

uh You showed us at the beginning the.

C

Yeah yeah: well, it's it's integrated in our code, but of course that could be uh exported as an uh library that would basically help you to do that. So that's uh that is a reusable code. Yeah, of course,.

B

So so we have here basically java classes within so so cxofloat is built as far as mentioned on on java springboot. So we have some essential java classes implementing everything which is mentioned here so yeah. I guess it's reusable in that sense.

A

Cool, I think we also have some projects somewhere that we wanted to open up uh regarding exposing some of those. uh You know as a java classes that somebody else can use it. Maybe we can pick it up from the shelf uh cool. Let's see if there are any other questions, uh no, nothing in the chat.

A

D

Have yeah uh so I've seen that you use this c as rf token uh I've. As far as I remember, we had some integration that is able to do that automatically. So you do not need to have that token handling uh manual. One. uh Have you checked that with uh the possibility or is missing uh uh registration of with the marketing? Yes? So this uh I think the application gateway from application connector supports that.

D

Sorry yeah. The question is: if, if, if you are not aware that it's there or you, it didn't work or.

B

D

B

We actually, to be honest, we didn't maybe dig too too much into that right. We wanted to get the scenario working and uh we saw that we need to propose, but we could share some some documentation. I think we could improve, because from two to one calls that would be uh very beneficial. Of course,.

D

Yeah, I I think that we have that support, but yeah, maybe uh yeah. uh I I will send you the documentation afterwards. So.

C

The integration I think it was created from time ago from my 1.99, I presume, and uh I'm not sure, maybe it's uh what changed or maybe I have missed it. Yeah.

D

I I have other questions yes, so uh uh you embedded compass in in keema, so you run it on with the compass uh deployed in your keema, runtime yeah and you use completely open source or for that or for this part I'm just yes. So uh how do you uh yeah like this deployment with compass and all this cooperation with compass? It was clear how to do that. I think you are the first one who is deploying the compass together with.

B

D

With keema uh on in open source.

C

Therefore, this yeah the application connector, which we touched uh initially but not in a big scale, and we saw that the camera is moving towards the compass direction and at least was written that the connector was to be duplicated soon, so mm-hmm open board already or in every stage to use compass, and I think it's worked pretty well for us, and this uh on me personally. I personally really like this graphql director, which uh compass has on this gateway and which helped us a lot in regarding of implementing these packages, these connections and api definitions.

C

So for me personally, uh there was no problem in using compass and it worked out really well.

D

Additional questions, so you mentioned something when you, when you registered this credential this basic credentials to your packages that you expected that you would so there was something missing that you wanted to have it, for instance, or yeah, can you can you clarify? Because maybe I didn't listen carefully.

C

Because we saw- or at least we had some hints from one of uh jamie jamie yeah- one of four guys yeah- that this could be actually used and I think, to have new credentials whenever a new instance for this package is created and we have per package and I- and we also wanted to have- have these credentials. For instance- and I see we see, this schema basically empty object and uh I'm kind of lost uh how to proceed from that.

C

So that's that's when I'm telling maybe some uh input for this hema would be a great help for us.

E

I can comment that as well, because I will have one question in comment about what you ask about the packages: how to organize apis and events. Yeah.

C

E

Pb, you will finish your questions. I can yeah.

D

Yeah, so I think that pk actually uh yeah sophia kopchinsky that that this started yeah. So I I think uh uh he and his team implemented the flow with adding these credentials. I do not know how we can do that yeah easily right now. Probably it's possible yeah to do that, for instance, already yeah, so.

E

Yeah, maybe I will comment. I will try to answer your first question about how to organize apis across packages. There is one feature with packages that, when you will bind many apis inside one package, when you will provision the instance, you will get one credential to reach all of them.

B

E

If your apis are designed that they could work with one credential, then it's the way to package them together. If they need the separate authentication types credentials, then you need to create separate packages. That means one api per package. From the eventing point of view, it doesn't matter it's just a group of apis, but it's related to the apis.

E

When you will need you will need to reach the credential, but more or less when you will create the service instance and activate 10 apis, you will have one binding right with 10 urls two different services, but with the same credential right then that's that's the answer about that and about those credentials. uh Yes exactly they are. uh When you will, we have such a feature in the packages which is called default instanceof and when you will provide that thing, that means all of the service instances will get copy of that credentials.

A

E

Means they are the same. If you leave default instance of empty, then application or user needs to provide credentials. That means in your application, when it's paired, you can add such a thing that you will monitor those uh uh off instances, uh requests in the package and then your application could provide one by one new credential and then in the service catalog. You will see okay in progress.

E

Okay, it's done created when you will delete the service instance, it will delete the it will, set the auth instance to status to be deleted and your application should remove the credentials on the application side right and more or less that flow is already there. You could just use it right in the application, then you have that feature. What you said that you you want to have credentials per service instance.

C

And this, uh and whenever we provision this new instance, that means that there's some uh call to our servers right.

E

Right now there is uh right: now you just have the option to pull information from the application point of view from compass. uh There is the some ideas about the push mechanism.

C

E

You will register application, you can add the callback url and you will will call you with some action. You need to do, but it's not implemented yet, but still you can use it and the the lessons I want to talk about that service instance has one credential and when you create service bindings, then it's a copy of that credential right. If you want to have new credentials, you need to create new service instance.

C

A

C

Guys this will be recorded.

A

So I guess you know you can go back pk. You had a lot of recommendations. uh Yeah.

D

So but this is why why I asked this question because I think that this is already implemented and you can you can use that. So it's more a matter of removing this default uh instance of from from this payload and monitoring the packages uh actually of instances uh in your application and then you can provide new credentials per instance. Yeah.

E

And in the next print, I think when you will use 1.14 the compass ui, we already have those options that you can set a default instance to to none. You can set default instance of when credential is known. That means all of them will be that if you have some, I don't know demo apis without any authentications needed. It will be the option to do that and we will. We will add to the ui much more uh tooltips and that's yeah. If you will do that. That means that and so on then.

C

E

Will be self-explanatory in explanatory in the in the ui itself, right.

D

E

You can always back to me.

D

So we promised you two enhancements yeah, so you can have credentials per instance and you can have uh crc is rf token handling. Also, so you can simplify the code. That should be very nice. Thank you. An.

E

Additional question: how did you install compass, you did it with the keema installer or you installed it as a separate instance.

C

We run the installer basically and it uh we. uh We have these override files, which we just appended compass to this installation package and each installed together.

E

Okay, because lately we removed compass experiment module from the kima itself, but you can go to the compass repository and there is the installation with those overrides and so on that we a bit separate those two things and it works that there are two installers right now: camera installer, uh kimaranti, star and compass installer right and you can play with that as well.

C

D

Good but general good cool stuff, guys, yes,.

A

Cool, um let's try to see if there are a couple of questions, also uh well one uh from sebastian. If he asks you if you're using the self-hosted or you're using the sap cloud platform, extension factory, keema runtime, which is that's the official name of the manager. First uh solutions: are you guys using the the self-hosted or are you guys using the the manage offering.

B

Yeah, that's the I think, mark also answered yeah, that's essentially the self-hosted. uh What we did in the past when this sifu hana cockpit was available. We provisioned in the beginning came out through that on the azure, but yeah not now, since it's like deprecated, we essentially self-hosted that, and we also got some insight like from uh marco and and from jamie about this uh managed.

E

B

So I guess this is the next thing we will definitely want to look into and then yeah see how it might fit. Also, our customers essentially cool at the moment it's a self-hosted uh in yeah in for.

C

Development we have this the most optimal solution for us and uh yeah. It's living on this heaviest server and connected to our systems, but for production then yeah. We will cross that bridge. When we come to it. I guess cool.

A

uh We I'm looking at the time we only have 20 minutes pb um is it? Is that enough for you to present the k3s or sure okay, so.

E

Then, let's try to understand.

A

It and then, if there are any questions we can bring it up at the end. Okay, thank you guys. It was really really nice to see, and I think for the team also was very good to see how some of our clients are struggling or who are using is using keyboard struggling. So uh I'm pretty sure that uh we will reach to you offline and we will give you the the full uh solutions or as the one that are coming on the pipeline to to optimize uh your uh your implementation.

A

Thank you for coming in and, let's see, if there are questions at the end, we can we can address them. Thank you.

D

Yeah, I'm ready, I'm probably already sharing. Yes, you can confirm that would be great, uh so uh just wanted to uh show you uh some repository I uh created uh actually last week and uh to play around with our poc uh with local installation of keema cluster on a k3s cluster. So this is simple repository that actually contains uh one script one chart, and so the essential script is schema.

D

K3Dsh disclaimer. It works right now only on mac os. So it's not supported in by other platforms on other platforms, but this is just experimental, but I encourage you to play around with that and give some feedback and we are still working on that and we will uh add other features here, but what it does yeah so uh so, in short, uh you need cube, ctl hand, free and k3d k3d is the uh the docker uh wrapper around k3s cluster k3s is uh analogy to k8s, so it's a kubernetes, but five letters less.

D

So it's just a slim down kubernetes, which offers on the essential apis. So all the predicated stuff is removed. Some uh some components are replaced by a more lightweight, for example, for example: etcd is replaced by uh sqlite and other things. So it's really um yeah a fast thing and uh what you, how you can do that you can run just schema, uh k3dsh uh command uh in your uh uh in your uh in this repository, and it starts this k3s cluster and installs uh keema components.

D

uh You have here on the screen, the uh quick version of that. So you do not wait uh all the time uh that what it's doing the installer is installing istio. First, then, all other components are running in parallel. 21 charts that are included in this installer are installed in parallel. You can see here on that flow that is taking some time. So it's about two minutes and a half on my macbook to install uh um istio and uh and k3s cluster and another two and a half minutes actually to get all other components running.

D

uh I will not uh show you uh live it's at the end. You get the url and password to the keema console, and you see on that. On that case it it took about uh five minutes.

D

uh I did it uh just before uh my presentation uh today, so I will switch to uh to my to my uh uh console and you see that I started started that just 15 minutes uh before uh this call to not waste your time. I will find the beginning here. So this is this kima k3d cluster and you see uh all this installation on my machine. Actually, it took to take to get this to almost three minutes and then entire installation. I will scroll down to to the bottom.

D

It took six minutes and 30 seconds because I was in other conference call and zoom took some of the power of my pros uh cpu. So uh without that it it takes a little bit less uh what you get yeah. So what you get is uh the url to console local keema pro, and if you go to that url uh I I will go there.

D

uh If you go to that url and you provide the password that is listed here, uh you will get just this.

D

You will get the access to the cluster and it's running in your local machine in docker you see additional namespace with search manager, so this installation also is issuing certificates. You see, I have trusted certificate here, so the lock is closed. uh It's okay and I can uh play around with uh for, for example, uh serverless functions. So let me create one um to prove that yeah, this stuff works really create function. uh Second hand function, that's bad, but yeah uh the function is building.

D

uh Let's wait a few seconds more and in that time I will show you that my docker is configured and right now build of function is executed, so for cpu for cpu is consumed. But then you can see the memory consumption on of entire keema cluster is about four gigabytes and for cpu in in the load.

D

When the building is finished, the load should uh calm down and I should have about 100 cpu uh consumed. Let me show you the resources, so uh I assigned four cpu and a gigabyte for for docker uh to uh to run that that thing and yeah. I think that the build is uh finished and also it's deployed. I can directly go to api rules and expose that service second hand.

E

D

Test just create, and if I run test not yet yeah hello world is already here. So you see I added new. I exposed that test in secure way. I didn't change anything in my etc hosts it's also covered by this local installation. So you do not need to take care uh of uh of anything inside uh your uh yeah change, any anything except adding this uh certificate to your uh trusted. Keychain uh the script is uh yeah. Maybe uh uh I will show you uh uh what it does.

D

uh So it's uh yeah uh setting couple of uh uh environment variables. You have local keema pro, which is the the the domain and this local keema pro points to uh 127.001, so it's localhost and your cluster actually is running on the localhost. So when you create the cluster here on the on this line, you publish 443, port and 8080 and 80 port on local machine, and this is the command to create that cluster.

D

The cluster starts up in about 10 seconds, uh so that's that's the time uh and then it goes to create all the namespaces required by keema install schema uh cluster essentials. So this is more mostly custom resources that we define and and eastern you see, this ampersand uh signs at the end, so this uh commands are executed in the background, so we can do that in parallel and then we have one command for waiting uh for the istio, so easter is installed in.

D

It depends on the power of your machine from two to three minutes, uh and then you have installation of all the uh charts we have in the resources folder of keema uh every of them in parallel. So so we have 21 jobs in the end that are running in parallel and then I'm waiting just for these jobs to be finished before I display the console url uh in the in the command line. So that's uh so average time is.

D

uh It depends, of course, of load of your machine as assigned resources, but I was able even wants to run it below five minutes, but yeah we are aiming to uh simplify some some some components uh have them faster right now uh or a hydra stack and and application connector takes uh most of the time. So this is the the longest installing components, so you can actually start already developing without waiting for them. So if you want just to have serverless, probably you can start even in three minutes from start yeah so uh and yeah.

D

You can have this script yeah. So if you do not want to have anything related to service catalog, if you are free to just yeah comment it out- and this is the way uh how I did, for example, with helen broker here, yeah, uh that's that's the thing and yeah questions. Maybe.

A

Maybe we should we should establish, because one of the questions from john is already uh if we recommend this instead of using mini cube. I think this is still a poc right. As you see you know, we have comments now, but it was a proof of concept to see if it works and if it maybe you can, you can answer that question if this is the official way of installing this locally.

D

Yeah, so uh so actually it's uh not recommended yet because this is just you see, uh but we will continue with that. So, uh as I said at the beginning, we will add support for that in the uh kima cli.

D

uh We want to do that, and there is already my colleague working on that cyan and we will yeah have also some uh uh ci tests for for that to prove that the stability and then we will also compare with times and uh and resources with mini cube, and then we will pick something probably so I expect that uh yeah, maybe mini cube, will be not our main uh way to install locally uh in the end but yeah. This is the early stage. That's why it's not replaced here. Yet what are the main problems?

D

Making this work on ubuntu? The main problem? Is nobody tried it, so I do not have ubuntu and if you want to try feel free, probably on linux. I do not expect too much. Problems should be rather straightforward, so yeah, there is no big magic in the script yeah. Maybe there is the set, but this is uh uh maybe doc docker inspect here, uh so I'm here uh using local uh local docker registry.

D

So this is another thing that I added to this to this poc when I maybe I will come back to that uh so, except of so, this is not only keema server. The cluster. I also have regis docker registry running on my localhost, and this docker registry is configured with this kubernetes cluster and this kubernetes cluster can use images from that registry. Why?

D

Because for local development, I do not want to push my images to docker hub or to gcr whatever I want to push them locally and then use them directly in in my cluster, so to have it here in uh yeah in the future. I want also to test the air gap solution so and the the way that we can use that I can have all keema required images in this registry. So when I start the cluster, it doesn't fetch a gigabyte of images from the internet. It's it has it all locally.

D

So then I expect I can also increase a little bit uh uh to decrease the little bit uh installation time. So that's uh that's the thing uh and so yeah from from the ubuntu point of view. uh uh Maybe this uh and docker uh inspect maybe will be different. So a different thing will be created here, but I do not know so. We need someone to uh to try and uh I have.

C

It so I can try it. I I see that this uh certificate or keychain part.

D

Oh yeah yeah this this part definitely will not work uh in in ubuntu so uh but but you can find easily the commands to uh to trust this. The certificate on linux yeah so.

C

I see that you have where this uh keema crt is uh so I guess when. If I went right in my own domain, I had to basically.

D

uh If you want to try your own domain, uh it's uh uh so it is so uh why I did did that yeah so and this local keema pro points to one uh 12701, so localhost yeah. uh You can do that with your own domain uh and I added something which is called ingress dns.

D

I will show you ingress dns search, so this is the chart where, where I added the request actually- and there is a self-issuer and you- I request this issuer from search manager and this installation is using server manager. So if you use your own domain, you can have certificate for that and you need to just add easter ingress gateway to your dns and that's that's it yeah. So you can also use that in this way, so you are free to uh to to change it yeah.

D

So I provided that this local keema product already points to some uh to your local host. So you can have this wildcard uh dns. So then you do not need to um modify etc hosts, but that's only the only thing but yeah the search manager is also here. So if you want, you can uh change issuer to uh let's encrypt and you can have even sign certificate uh if, with your own domain, uh if you want yeah so cool but yeah, this is not uh covered yet by by this.

D

So on k3s I uh focus mainly on the local development. So that's that's the thing. Another question.

A

Do we see any improvements in the in the you know running on your machine? We saw powell's and pawels. You know struggling running the the instance of their on their computer, so yeah installation is faster, but what about uh after installation while it runs, does it have any.

D

uh Yeah so yeah, so I can show the the hyperkit. uh When you do something is about 100 to 400 percent of cpu. uh It depends uh sometimes it's it's doing something uh I'm still investigating what's happening so which component in idle state is generating some load.

D

I've seen that we have some certificates, changes refreshing whatever jobs, I'm just hunting down this, these jobs that are causing some some load, but you can see that already from the dashboard that that not this one, this one is real nice and that it's uh four gigabyte and yeah about one cpu and I'm still able to do something on that computer. I am able to join zoom meeting and yeah and build even some lambda. So it's not so bad uh and yeah.

D

My fan uh of of of my my book is not working on the highest uh uh yeah spin uh yeah comment.

B

Do you have now on your macbook, so how many like cpus, you have.

D

So it's it's a cpu yeah! So it's it's! My macbook yeah it's on two years old, macbook pro yeah! So.

A

Oh, it looks promising. Have you already seen your notification that you have a new meeting in two minutes so and we are getting close to to the end of the of the uh meet up? uh I just wanted to ask if, um if there are any other questions, maybe for pavel and powell's as if they are our guests, maybe there are some missing questions, but.

A

We are waiting for them in the chat. If nobody has then I'll. Take the opportunity to sorry supporter.

D

And paul, please stay in contact on our keema chat, slack channel yeah, so we we can talk uh yeah, especially with uh you, should be also paired with uh kopchenski. That's uh to proceed with this uh credentials. Yes, uh I.

B

Think both of us actually on the slack uh community.

C

So I already have some have had some questions uh resolved by you, so so it's uh seems to be working, so we will continue stay on slack and.

A

Oh we're glad to hear this. That's our main support. uh You know, no, no commitment, you know best efforts, type of support we do have uh support guaranteed, but for the management yeah for sure.

B

But just in case I also share my email in in in chat, so if uh yeah can easily reach us mine and pause basically, but but we are also on the slide channel. As I mentioned.

A

I will add all the links also to the to the keymap community meetup document, so we know also pb from your site where people can contribute and also um maybe help out with uh pause. If you want to help out with ubuntu, you know, I think we will be more than happy to to take that. It's still.

D

Appealing requests are welcome, yeah, so yeah. I I yeah. So it's my favorite project right now. So if you do the pull request, I will review it in five minutes. Five minutes yeah.

A

And next time running on raspberry pi's right tv.

D

Yeah but yeah, we need to call, we need to compile easter, so it's not so easy. It should.

A

D

A

D

A

If this runs on raspberry pi, then I think everyone can run it on their laptops cool. Then uh I think that's it. Thank you guys very much for participating uh presenting your work, I think was amazing, was really appreciated and also was insightful for us to see. You know what are some of the challenges come back on slack and we will try to continue to support you. There uh pb was very nice to see this running.

A

We will hope to to see this uh faster running on our laptops and hopefully we'll make it to uh to the main cli or as an official. Well for now it's it's something that everyone can try on their laptops. So yeah with this in mind, see you in two weeks guys. Thank you very much.

D