Red Hat OpenShift OpenShift Coffee Break | Red Hat Livestreaming, 1 Dec 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: OpenShift Coffee Break: Open Cluster Management

Description

Get your espresso ready for the EMEA OpenShift Coffee Break together with Natale Vinto, Jaafar Chraibi and Tero Ahonen as we welcome our guest Gianni Salinetti, Solution Architect at Red Hat, to celebrate Open Cluster Management’s acceptance into the CNCF Sandbox and discuss all about multi-cluster multi-cloud Kubernetes deployments!

Twitch: https://red.ht/twitch

A

A

B

Good morning, good morning, everyone welcome to this other session of the open shift coffee break here at open shift tv uh with this show uh about everything about openshift and cloud native architecture. uh My name is natalie vinto, I'm product marketing manager for openshift, and I'm here with my colleagues and mate jafar, hey jafar. How are you.

A

Hey good morning, everyone uh it's great to be back here and have some coffee, although I forgot to take one this morning. So let's uh yeah, let's, let's see how it goes.

B

Go remediate immediately because today.

A

We have a very.

B

Hot topic, which is the open cluster management and I'm very pleased to uh welcome our guest today, which is uh johnny salinetti solution architect. I read that hey johnny.

C

Good morning, good morning, natalie here too nice to meet you happy to be here so uh today we're gonna talk about open cluster management, submission on cncf, yeah, very nice, very nice topic.

B

I really looking forward to it johnny because uh um I'm very interested on knowing what is really this open cluster management. How is related to kubernetes multi-cloud multi-cluster management? I know it's a hot topic in the market. Everyone has its own solution. We also have in red that openshift our solution, which is acm so jenny, correct things are wrong. This is the upstream. Those are the upstream bits of advanced cluster manager, management for kubernetes.

C

Yes, correct ocm or open cluster management, uh it's the upstream for red that balanced cluster management for kubernetes and um it's totally open source.

C

So one important thing is that red dot acquired the solution uh some time ago and open sourced and everything was open sourced uh following the usual behavior of red hats, in this case, so where red hat uh gets uh acquires so a product, uh maybe a property software, and it becomes something that that is managed by red hat. It's uh open sourced maybe takes some times.

C

Maybe it can take two months one year, it's up to the how it's written but fine, it's open, sourced and ocm open customer management is an open source, uh completely usable version of uh so of the multi-cluster management solution of reddit, and it's the upstream that drives all the changes to icm and the nice thing. Very nice thing is that it's being currently admitted as a sandbox project to the cncf.

C

So that's something we can discuss later, because it's very interesting because it it's a very important signal to me. uh It means that this is something that it's really ready for the community.

A

C

A

That's great and uh yeah correct me also, if I'm wrong, but uh what I find interesting with this specific uh product from the acm standpoint and project from the community standpoint is that uh we all know that red hat has been acquired by ibm some time ago, and maybe some people started.

A

Freaking out uh is red hat going to remain committed to open source and are they uh going to hold on to these to these core values and, and things like that, and if we can also tell the backstory uh the the the product itself comes from ibm engineering.

A

In the beginning, it was called mcm, multi-cluster management for kubernetes, actually and uh after the acquisition, the engineering was transferred from ibm to red hat and not like the other way around from red hat to ibm.

A

So I I think that was a very cool proof of uh both ibms and red hat's dedication to committing to open source, uh and now we see the the the outcome of it, which is uh we have the red hat product, which is red, hat acm and we uh created the upstream and the community for it uh with open cluster management. So I think it's really um satisfying to see that red hat stands to these values and and see that we are still you know, pushing things in the upstream.

C

I I wanna point out something: that's very important uh use, you know. Sometimes a project can be open-sourced, but maybe the upstream maybe cannot be fully usable. So maybe you have a project many. You have many repositories on github, but it's hard for people to assemble stuff make it.

A

C

A

C

It's here it's that open cluster management is something that works out of the box. So it's a distribution that works for everybody on every kind of kubernetes cluster. It can be installed. We now showcase for on a basic, vanilla, cluster, certain kind, so very basic basic thing, not an office shift, it works everywhere and it's something that works out of the box.

C

A

B

A

C

Obviously, have your edges your repositories, you can investigate the code. You can contribute to the code very important, but you have something usable.

A

Yeah, you don't have to rebuild the whole thing yourself and and be like a black best engineer to to to have something ready. You have the binaries there and you have the container images and, and it just works. So that's that's great in terms of.

B

Just making it.

A

Easy to to adopt and use.

B

Yeah- and I I like what you said jafar- that the the stories uh uh they read that way, you know to do the things an acquisition, then the open sourcing of a product and.

C

B

We see the results, so I would like to share in the chat uh the the the website of of the result, which is the website of the opel open cluster management.

B

If you go to this website, you will see all the details of the project, all the github repository and how to join this community, which is the very important part of uh also of the the overall process, and I also would like to ask johnny before we go into. I know we, of course we have a live demo right.

B

Of course, we have uh demos and where johnny will show us how it works in a local kubernetes and how to to make it uh working with multi-cluster, uh but before we go in that johnny, what that means that a project is uh accepted as a cncf sandbox project. What does it mean for what does it mean for for the community.

C

It's very important because it means that the cloud native.

C

Yeah, sorry, uh the club community foundation is ready to include that project in all their ecosystem. So it's it's included in their ecosystem map. They have many degrees of acceptance, so you have sandboxed you upgraded, graduated and match your projects, so you start from sandbox and then you graduate so it means that the cncf is looking at. This project is uh helping it grow and it's a huge up. It's a huge boost, because when something goes inside this cncf sandbox, it gains a very important visibility and boost.

C

So it means that we are something that is going to grow. It's going to grow quickly, probably because we have something that it's ready to be used. It's ready to be adopted by a different kind of kubernetes users. Obviously it works great in openshift, because you know we are talking about openshift after all here so, but we have something that it's not difficult for kubernetes, so it's just worth using custom resources, so you can apply them everywhere.

C

So the point is a sandboxing project is something that is really helped in growth. That's my my opinion about the importance of being graduated to the cncf.

A

Yeah, but I I, I also think that it means that there is acceptance from a broader set of users and contributors than like, I would say, the traditional red hat fellows, because cncf is a broad scope of I mean it's a very large ecosystem right and being sandboxed there.

A

That means that there are people that adhere to like the the the the tool and the features that it can provide and that this can be useful for kubernetes, as as a community and as an ecosystem, and not just like an end software vendor solely uh which is red hat. So that's uh also, you know this this notion of being accepted and that it serves uh a broader purpose uh for kubernetes community itself.

C

It's something that is accepted somehow by the community.

B

We here we hear your microphone a little bit far. I don't know yes.

B

No, no, I am probably some some volume setting.

A

C

Heard your keyboard.

B

That we know that you are prepping uh super cool stuff to show us, but we also see you less. Maybe.

C

Maybe now it's better yeah.

B

Yes, all right.

C

Thanks thanks all right, maybe I can try to show something natalie. What do you think.

B

Yes, uh uh before you go, I would like to um to ask the people attending uh so there's a tiger flight that say that the microphone was not good, but I I I think now johnny solved it. uh I we I hear a little bit of echo slide the echo, but um I think uh it's fine. So now the microphone is fine and I would like to ask to the people attending if you have any question about uh open cluster management on everything. Please uh write in the chat.

A

B

Will bring the question to to the speakers uh during the live.

A

So I I do have a question if, uh if I may so what what? What is this? Actually, what is it and what? uh What is the purpose of this tool like if you can very briefly, introduce the purpose of it, its main functionalities, uh I'd say even before going into the demonstration yeah.

C

Absolutely absolutely right: uh the point is managing multi multiple is complex. You know when you have to deal with many many many clusters, it's complex to manage resources, keep them aligned and checked so multi cluster management is a topic. That is a long story. It's probably it's something that has been discussed since the beginning of kubernetes, and there were many solutions.

C

There were many attempts to manage this. um The concept is, you need a single pane of class to manage all your clusters. You need a single point when you can apply your changes uh where you can monitor the state of your clusters when you can monitor the parameters for your applications and also maybe when you can manage in a declarative way, your infrastructure declarative way. That's a very important point.

C

You need to, for example, to manage your application lifecycle in this way, so maybe using a githubs approach using a git repository when you have your your applications configurations. So all this stuff together and also you, maybe you need to manage policies for your clusters, also in a declarative way. You need to integrate policies as soon as you join your clusters. To your to your multicluster infrastructure, so.

A

C

These things together can be solved by a unique solution which is up in cluster management without.

B

The cluster management- you essentially.

C

Have one manager it's called hub cluster? Usually that's it's responsible to manage or a series of clusters, it's responsible to apply a series of custom resources to your managed clusters and you can do it uh in many ways. We're gonna see how and it's responsible to monitor the state of your clusters, which is very important because um you can see the the health of your clusters. Since you have agents running up your on your managed clusters, which can continuously report the status of your cluster and using those agents, you can apply changes to those clusters.

C

You can make groups of clusters, for example, you can mage manage cluster sets and apply in embark changes to a single set, and uh you can you cannot? You can do a lot of things and one very important is application life cycle management. Maybe if you have sometimes later, we can try to also to apply and add on the application life cycle at dawn and see how you also can do heat ups with open cluster management.

C

A

C

Declare your application configuration on github and do it and apply changes on um on managed clusters. So all this scenario makes the um use case. The basic use case of open cluster management and.

A

C

To do this on rcm, you have two main components: uh a cluster manager running on your hub cluster, so you need a central hub cluster to control all the others, uh just like the lord of the rings wearing to room at all to rule them all. You have one hub cluster to root them all, and you have managed clusters which run those agents called cluster. Let's the cluster, let's think of it like um like cubelet, in a work in a kubernetes worker node, the cluster.

C

Let is the agent it's an operator which communicates with the hub reports the status of the cluster, its health, its metadata and also gets from from the hub cluster informations about the changes that must be applied to the managed cluster. So you have a work agent in there in a managed cluster. That continuously enforces it's reconciles using a kubernetes related term the changes on a managed cluster. So it's it's a matter of reconciliation.

C

You know it's just a matter of reconciling in a declarative way, so you are just doing say doing the same on the manage clusters. That's the main concept of of ocm.

C

um Also, if you want to for people who try to understand it better, if you go to the uh ocm documentation, which is on a link that natal shirt, you are, you also have nice, uh nice, graphs, architectural graphs, about uh the design of ocm. This is the basic design.

C

Then, on top of it, you can add many addons, for example, the application lifecycle management, which is an add-on. You can add the policy manager which is another add-on who lets you deploy policies on your clusters, so it's some kind of some kind of uh configuration management, for example. You maybe want to apply default configurations.

C

Maybe you want to create some config maps on your clusters as soon as they join your your your cluster set. So it's great for that purpose. For example, you know in acm, so you know open shift downstream related.

C

I use these policies to deploy operators. As soon as I join a cluster to acm. I apply a policy that deploys one or many operators on my cluster, so without using uh is any other tool. I just need to join the cluster as soon as I join the clustering, and I apply some labels to this to that cluster. That match my policies.

C

It gets all the configuration immediately and that's great because.

A

C

Sorry, but completely.

A

Yes, I was going to say so some uh like uh uh examples that maybe people can can more easily understand is say you have new clusters that you created the 15 new clusters and you want them all to integrate to your ldap or your your uh central authentication system, so your users can can can have access without needing to to push the configuration to each one of them.

A

That's something that you can enforce like, say: okay, whenever we create a new cluster, just uh configure yourself to authenticate with our central uh authority or whatever tool we use, and that that's one step uh that you know makes it easier for your for you and your users and maybe something else like central logs. You want all your clusters to push backlogs to uh your central login system.

A

Okay, just push a policy there and whenever you have a new cluster, it's going to configure itself to send back the logs to whatever meaningful log system needs it needs so yeah. That's some types of like uh standard policies that you could push.

A

uh Of course, there are also some functional policies, like um I don't know things like pci or stuff, like that, where you can enforce some.

B

Specific types of.

A

Yeah some specific types of of policies, if.

B

You recall uh jafar, we had the demo from uh fuzz and andres uh in on our one of the previous show. They did a great demo about acm now, johnny, I'm really now. uh I would like to say this show me the code dude so.

A

B

Nice words now show us the code show us the evidence.

A

C

um I did this uh short deployment on purpose similar to the documentation. So if you look at the documentation, there's a quick start, which is very nice- and it's based on kind, so you don't need oh wow, big kubernetes clusters. No, you just need kind. What is kind kind is a tool for deploying kubernetes in that's the name kind. So it's an acronym or kubernetes in docker, so kind lets you create very small kubernetes cluster inside a container.

C

We are not gonna use docker, for this kind also supports uh podman, which is our our container engine. It's great it's great tool. It's dim or less it's great to use. It works in the router's mode. I use it in routers mode for this demo, for example. So you don't need.

A

C

Running podman, it's amazing, and so we run two cam clusters, uh one hub and one managed cluster, and then we deploy and cluster management on top, it's very easy. It's something that everybody can reproduce immediately right now. So let's share this video. Can you see it yep? Yes, yeah. Let's start it so now.

B

C

See I have two tabs here. I did one tab for the up and one tab for the manage cluster the manage cluster. It's going to be uh created right now. Now, I'm going to I creating the kind hub the ken cluster naming hub so kind is creating the cluster. Now I'm creating the cluster one. The managed see that kind says that I'm using the podman provider here, not docker, so it takes very little just a minute to boot up to bootstrap to clusters.

B

One question: if to enable the podband provider, did you provide any embodiment variable how to switch from a docker to a podman provider.

C

B

C

B

C

So now uh my cluster is started, so I just need to monitor if the containers are completely.

C

Let's see, I passed the context uh to use the specific kubernetes context here. Kind hub is the name of the cluster, so you see it's very fast. It's very small, it's minimal cluster. Everybody can run it on their laptops. It's very easy and it's great to experiment with kubernetes. It's a great way to to to play to do your first steps if you are starting to learn kubernetes and also to make some quick experiments.

C

Okay, everything is started on hubla. I just want now to deploy stuff. It should okay cluster again. What's this, it's the cli tool that uh deploys the open cluster management. I pre-installed this cli. Obviously so just one comment: cluster idm in it it each initialize my cluster, my hub and uh creates this message. This is the uh command. With the token I should run on the manage clusters to join them now before joining the cluster. Let's see if the deployment of ocm has completed, so you see it's still creating the containers.

C

It has created two new spaces: open cluster management and open cluster management hub which uh contain the cluster manager and also the other addons.

C

As soon as as it's finished, we can move to the manage cluster.

C

So you see it's very easy. I just created the hub cluster. Nothing more just one comment: that's.

A

So yeah the this might be obvious, but uh all the components are actually containerized right, all the so it's it's a sort of micro services based architecture for ocm and acm and all of them run into containerized.

C

Yeah, absolutely absolutely so now I I copy the join command and run it on top of the manage cluster.

C

So I just passed the cluster name and the context, so I want to run this on top of the manage cluster because I need to say to the manage cluster a join the hub and this. This is the token to join the hub.

B

Very easy: that's cool, so you basically have two kubernetes nodes created by a kind in your workstation, so anyone in linux, mac windows can install kind and reproduce the same demo and you are joining those two cluster with the one. Is the control control the control plane and one is that the con there? It's a data plane or a control led cluster.

A

Manager, yeah yeah.

B

That's cool, so we have two nodes, one is one is controlling the other and just one two.

A

B

And just one command to join one cluster to the to the manager. One.

C

Yeah absolutely easy answer. The the the team of ocm did a great job with the cli, so you have now uh just a message to jo to finally accept the cluster. The manage cluster now made a request to the hub. Hey, please join me, so you have.

C

uh You can see that you have custom resources on the hub. You have this very important custom resource the manage clusters. You can see that you.

B

C

Appending manage cluster, you see hub accepted, false is not already accepted. The hub must accept it. It's just a matter of accepting a csr here. It is. We have this pending csr.

C

So by running the cluster adm accept command. We implicitly uh under the hood. We are accepting a csr. We, we don't need to know this stuff. Indeed, I'm just showing it to make it more complete, but you see I just run crosstrading accept and the cluster is joined, so everything.

A

C

Please please jafar.

A

Yeah yeah: this is interesting because you, the the the the managed cluster, had nothing uh had no relationship with the with the other cluster beforehand, so it wasn't created by something uh else. It's just like a regular kubernetes cluster.

A

Once you execute that command, it deploys like the cluster leds and whatever it needs, and then it it reports back to the hub cluster. Is that correct, yeah.

C

Correct exactly.

A

So, basically you you import any kubernetes cluster as a managed cluster in there.

C

Absolutely any kind of kubernetes.

A

C

So now we can move to the let's stop this and let's move to the to a live shell, so live terminal.

C

Can you see my terminal.

A

C

Okay, so I want to show you something more inside the clusters. This is the hub, so you see now inside the app we have this container. Those containers here that are related to the cluster manager and if I move to the manage cluster and I run oc, get pods dash a dash context, kind cluster, one. I passed the context of the manage cluster.

C

We can also use I'm using rc. I can use cube ctr here. Obviously yeah you see, we have the twin spaces, open cluster management and open cluster management agent, and we have this spot deployed.

C

What's this the cluster, let is the agent that communicates with the control plane with the cluster manager on the hub, and the cluster led also deploys some extra agents, in this case the registration agent, which is the the purpose of registering the cluster and the work agent, which probably is to apply the changes that you declarative, uh declarative, define on the hub. So every change you do on the hub.

C

Every resource you create on the app is applied by the cluster led and its work agent on the manage clusters. So the point is you don't have to touch your many clusters, your managed clusters.

B

C

You can do anything from the hub right now. If you have one ten or a thousand, manage or manage clusters, it's just the same.

C

What do you think about this.

A

I think this is great, so one question that uh one might ask is: is there a big overhead on the manage cluster like? Does the these agents weigh a lot in terms of cpu ram, or is it something negligible.

C

Oh well, this uh this configuration here is quite lightweight and uh it's very basic, so the components we are running here are not using a huge amount of of resources. uh What changes is the api pressure as you grow your managed cluster number. So, as you grow number of clusters, you join to the hub.

C

You, you obviously uh increment the number of api calls between the cluster manager and the clustermint.

C

So this is something that, as an input on eip on the api pressure, what's the concept of api pressure, we can define it as a the saturation of the api calls on your cluster, so your hub, a a limit of api pressure that it can stand uh according to the sizing on your app. So when you, when you design the sides of your hub, you must be.

C

You must be aware of your of the number of your managed clusters.

A

Yeah, so that's that's why uh the hub is running on a dedicated cluster. I guess so. It can scale up to manage thousands of clusters and I believe, uh on the acm parts. We did some tests up to 2000 managed clusters or something like that, which is quite quite a lot.

C

So the point is as much as you go as you plan to grow your infrastructure, you have to somehow adapt the sizing of your of your app cluster. uh Obviously, when we deal with a small number of managed clusters.

B

C

Have just one: it's really uh really: it's more small small resource usage.

A

So here everything runs on on your laptop. Is that right.

C

Yeah I'm a laptop alongside with another kubernetes cluster running on virtual machine. My chrome browser with 40 tabs opened uh one one.

A

C

For plain music hardware, which is a dual software for linux, so I think that my.

A

C

Managing it very well.

A

Very nice completely lightweight from what we can see. Yeah.

B

Very lightweight, I have a service communication. Unfortunately, we cannot stream uh at full luck hd at the moment, so from youtube. You would see 480 uh because the I think there's a wrong setting in the stream machine we are using. We will correct in the next show, but for now I'm sorry we can have full hd. I hope you can see everything uh enough good, so agiani to help this. Maybe you can increase the phone. That's the maximum.

C

Because the people cannot make.

B

It full full hd, so thanks and sorry for this problem, no.

C

Don't worry, um can you see this font size? It's is it readable? Please tell us.

A

Yes, I think it's uh it's uh easier to read now: cool.

B

A

Should be fine, okay, all right so show us the magic. What can we? What else can we do.

C

Yeah first, you created the manage cluster. So on your app, you have a resource, that's it that is a managed cluster resource, and you see the name of your manage cluster cluster one so youtube. It is a custom resource, so you can do it. Something like this.

A

So so, just for people who maybe are not that familiar with kubernetes basically a custom resource is an extension of kubernetes api to add your own concepts that become a native resources of kubernetes. So now, kubernetes understands the notion of what a managed cluster is uh because we extended the api to uh through the ocm um installation. Thank you.

C

Thank you all right. I missed this. It was a very important point. Thank you.

B

C

So you see, this is the custom resource of the manage or your managed clusters, and you can create uh metadata on this manage cluster. You can apply labels, for example, and you can use labels to apply configurations on clusters that match a specific label.

C

I did also did a repository. I can share it in chat with some extra examples taken from this session. I share it on chats there. It is, and here we can see, for example, how to apply a specific resource, a manifest work uh on top of the of the cluster. What is a manifest work? It's a custom.

A

C

As jafar just explained, that says to ocm, create this object on specific clusters so to create a manifest work. It's very easy. I just need to apply a resource. You know in a namespace named after the manage cluster. So if I look at my uh ocm here and.

C

Kind hub, I can see that it has created a namespace with the name of the manage cluster. Very easy. If I create something inside here, this object will be applied to the manage cluster very easy. So I can move to my repository.

C

uh If I move to cluster management directory manifest work, I have this file a low work. It's just a very basic deployment. You see very, very easy. It creates a deployment object inside a cluster.

C

So what happens when I create this object when I create this object inside the namespace cluster, one.

C

The the manage cluster, the cluster, let applies using the work agent, this object to the manage cluster and I just do it declaratively inside one directory in the hub. I don't touch the manage cluster.

C

Any any any questions about this is: uh is it.

B

No johnny jen, just one question: are you told that you're sharing the chat something, but we haven't seen anything? What did.

C

A

C

B

A

On telegram or on our internal.

B

Channel and uh in the telegram coffee break chat, I will put put it now in the in the in the chat so examples, repo.

B

Okay, now we should have all in youtube and twitch. Please let me know if you can see.

B

C

A

B

Examples johnny and can run the same example you you're doing uh right now, yeah.

C

Absolutely they cannot be, they can be applied on the manage clusters, as I'm doing so. If I create this resource here, I just created this resource inside a namespace named after the manage cluster cluster one. So what I expect to see now is here: it is hello, you see I'm on the manage cluster here and it just created the resource immediately on a managed cluster quite simple, and this is something that can be applied to any kind of resource.

C

I just created a deployment, but I can do anything just like jafar said before, and uh it's very basic, very basic approach. We can make more. We can evolve this approach, for example using application lifecycle management. Just we in that way we can enable a github approach, so we need to deploy some more add-ons. You find those add-ons in the in the documentation of the open cluster management.

C

So if you look at the docs, you see uh that in the architecture add-ons you see um the addition how it don't work first, so you see all the add-on manager you're deploying a dom manager on a hub. The add-on manager manages all the additional components you deploy and you deploy agents on the manage clusters that communicate with the dots.

C

So if I want to, for example, to to add the the application of segment management, I just need to create uh these components on the album manage clusters that are able to sense changes to uh repository.

C

So you can use git repos, but you also can monitor and charts or object stores, not only git repositories and using this approach you can apply changes of an application to the managed clusters and do it everything in a declarative way.

C

The nice approach here is that you start uh you start creating your application using the github's approach by applying labels to your clusters. So you don't need anymore to know, for example, the exact name of the manage cluster. You where you want to apply your your resource. You just say I want my application on all the clusters with the label: environment, lab environment prod, so every cluster with that label automatically gets the application, because you just declaratively declaratively created a placement rule. That's the name of the approach.

C

You define a rule that says how to place your application on your clusters, so in in the repo. I did an example as more example about about application lifecycle, management of a simple, very simple, nginx application, it's very basic.

C

So first, I'm going to show you how to install the add-on, and then we try to deploy the application in a managed cluster and see what happens.

A

C

Okay, first answers.

A

Yeah, sorry, so just um maybe to touch upon something that is very important is when you come to that scale, where we are managing uh tens or hundreds of clusters and does thousands of applications. uh One of the questions that arise is how do you control what happens where, like in case, there's a problem? There's an issue you need to troubleshoot.

A

How do you figure out what happened when what what configuration was pushed to what cluster? And if you do it like uh through command lines, I mean that's totally, something that you can do uh you you do a cube city ctl apply to 10 namespaces and it goes there. It's going to be deployed from the hub to your 10 clusters. That's fine!

A

But when you want to audit like what version of the deployment was deployed and when etc, it becomes a bit harder, and I believe that's where the getups approach becomes very handy is because you have a traceability of what was applied when to which clusters etc. So, yeah, I think combining the two uh is- is really a great fit because uh you.

B

A

The of course, the power horse of the hub and the cluster legs to deploy stuff, and you have the uh single uh place of truth, which is you push your one change to your good people and then everybody else gets it. But you know when and why etc, or at least when not necessarily why? If there are no comments but.

C

Yeah, I totally agree. I totally agree jafar, because the hard part is when you manage not to five but hundreds of clusters, and you need to keep control of everything. Keep it on. Keep everything under your control and know what's failing where and why.

A

And you said: there's a reconciliation is that correct also like uh if I go to one of the target clusters and tamper with the uh something? uh Is it going to be remediated yeah.

C

Exactly exactly that's a very important point: if you apply a configuration, so let's say I want to deploy an application to my clusters. Let's say I want to deploy this basic nginx application.

A

C

Say that somebody changes my deployment on the cluster using the cli. It goes there and scans out. For example, my application using cli so interacting directly with the manage cluster agents in the in manage cluster, reconcile immediately the application to the desired state and that's one key point of gitops. Reconciling constantly the desired state of an application inside a git repo.

A

C

Current state in the cluster, so every change that is done in a cluster without so without this approach. By bypassing this approach, you.

A

C

Clear concise every change you do in your clusters must.

A

C

Through the git repository, that's great.

A

So yeah very nice uh and, of course, very useful uh in in real life when you have to manage up to that scale. As we said, cool.

C

You know you know uh my sense. Many people love github's approach because also you, you can also keep control of changes. You keep track. Sorry often of changes. You see every single con. You can see every single comment. You can approve your commits because.

A

C

You can create.

A

Full requests yeah.

C

So you really keep.

B

A

B

Mentioning uh gitops, but I'm wondering uh if uh gitops means connecting to an existing github engine like argo, cd or any other like that, or is there any other github's implementation there.

C

Ocm has a native graph implementation with the application lifecycle management, but the the great thing is that you can integrate it with other tools like cargo city and that's exactly what we do with acm advanced cluster management, where you have a native ui that integrates with argosy resources. So you can see live what you create on top on your argo cds. So in that way you can kind of separate your infrastructure resources. You want to create, for example, with acm with your application resources you manage directly on the cluster with arbor cd.

C

uh If we have some time, maybe I I can show you an example of my lab, where, where I run uh argo cd resource is integrated with acm. But yes, the answer is yes, you have native github's approach and you can integrate uh with uh with argo cdo data tools, and so you know uh get the best of everything.

B

That's super cool, it's a agnostic! Basically, you can plug in as an as an interface right, so that makes really really powerful.

A

So I I I I had one question, but I I think you wanted to show something before I mentioned the get up thing. uh uh I don't remember what what it was you were. You were good.

C

Yeah, I wanted to show you exactly the application life cycle management, which is the uh step zero of the github's approach here.

A

C

We just uh deploy those components, this deploy those agents and then try to deploy in a declarative way, using github's approach, just simple application, and then we we can extend the concept with our cdn more stuff, but let's start with the basics. So now I can. I can show you that I I just create on my manage cluster, a simple comment with a simple command. I just want to kind hop. I install an add-on the net, the name of the add-on is application manager, and I used to do it with custodian.

C

The the cli of ocm, the great cli of asean. So just run this comment, so it's a stunning stuff. We can monitor what's happening.

C

Let's get you see, it's deploying more components. It's deploying some operators, these ones and these components are basically four. So you have a channel uh this, a component that manages channels a channel is the, let me say, a pointer to a repository. So it points to a specific git repository that works with our application.

C

Then you have placement rules that are the rules that match your managed clusters. So placement rules can manage conditions. For example, the existence of a label and subscription which connect together placement rules and channels and say, okay, this channel pointed here, which is a git repository, must be handled for these clusters here managed by the placement rule. So in this way you connect together uh the source of information, so the git repository pointed by the channel with the target, which are the clusters managed by the placement. True, so.

B

It's very easy.

C

And we have the project components, so we need to make a last step. We need to deploy the agent on the manage clusters. So to do this we run just a second command. We enable the add-on on the manage cluster.

C

So I want to enable this on cluster one. I do I'm running. um Look I'm running everything from the hub, I'm not going to the to the manage cluster.

C

Just enabled the addon on the manage clusters so on the manage cluster now as an agent as another container, another pod, which is going to be deployed on a managed cluster which senses the changes that we do on the application life cycle management on the app. So we extended in a very simple way, our hub with.

A

A new add-on, the.

C

Application life cycle management very easy, very, very plain and just comments just comment to enable yeah, john and just another command, to enable the addon on the manage cluster. Stop. We can do this with other addons, for example, policies.

C

We don't. I think we don't have time to show policies too, but that's fine, so I just want around to run the application. So if I move to my repository, I have an example of an nginx application.

C

I have resources that should be managed inside my cluster, so you see, I have objects. I have a namespace that is expected to be created on manage clusters. I have the channel, which is a pointer to my git repository containing my application resources.

C

I have a placement rule which says is this is most the most.

B

C

Thing which is saying apply, these changes apply these resources on all the clusters with the label and lab it's just label. I just create this label on the manage cluster and see it is applied. I can apply this label to a thousand of clusters and add the application deploy to a thousand of clusters. Just with one comment, and then I have a subscription subscription connects together of everything. So so it subscribes to my uh repository and applies the changes to the cluster, matched by the placement rule and inside the resources directory.

C

I have the files I want to deploy inside the cluster. The.

B

C

Just a deployment and a service very easy: it's a deployment of an nginx application, basic basic thing and.

B

uh In a service.

C

And- uh and everything is declared in simon, it's repository so.

B

C

Just want to create these things, but I'm missing something. I need to label my clusters, so I saw that my label was managed by the placement rule. I need to apply a label called envelope, so I just need to.

C

Edit, my cluster.

B

And it's it: this is all live demo. Folks all live demos.

C

Or live yeah, I just go here.

C

Context, sorry typo and you see I have labels. I just add a very simple label here and lab just like by placement rule.

C

So now, if I apply this resources in here, let's see what happens: okay apply chef dot. I just supplied the whole directory.

C

Okay, I have created those resources inside my hub cluster and, let's see what happens on the managed.

B

See a container creating.

C

Yeah there it is nginx up, you see, the namespace that's been created. I have my deployment created and I also have.

C

C

That's the that's the part and that's the service. Everything was created immediately and if I change such now, if I change something on my git repo, let's say I scale out my deployment. This is immediately reconciled to the managed clusters, that's great and that's exactly how it works in uh advanced cluster management for kubernetes.

C

The main difference is that there I also have a very easy, a very user-friendly ui to do everything not only the custom resources, but if I, if I want, I just, can use the command line and the the cli just like I did here. That's.

B

Cool, so the the core is open, the core is all upstream. Then, of course, the product has some ui, some things easy to install and maintain, uh and that is the supported, but the core is open source. The same bits are the one that letting you right now uh deploying nginx to two clusters.

B

uh The same would be an acm, because the core is open.

C

Absolutely and since it's open my my messages contribute not not just use contribute to the to the community, because this is a great project that can grow even more in many ways in my opinion and uh so the contribution to the community. It's very important.

B

That's great statement: I'm gonna put in the chat, contribute and there's the link in the in the website of open cluster management about contribute how to contribute how to join the community. On github on the working group on discussion about this new community, I'm really looking forward to see that community growing more and more in the kubernetes and cloud native space.

A

And just keep in mind that contribution is not necessarily by pushing code. You can also contribute to things like maybe documentation or samples. We saw that you just used some samples that you have on your repo and if people start to build interesting things like that and then share share those on on the public repos. That can be great. Also, you know to help uh all of us.

A

um You use those resources and touch things so usually when people are here contribute to open source, they think. Okay, I have to be a black developer or something like that, but it can be different. Things.

A

Contribution is also like filing bugs. If you find bugs you file an issue on github and you contribute to make it better.

C

I think that a great starting point in contribution is helping improve documentation. For example, you don't need to be a black belt developer to contribute to the limitation, and you do a great huge work for everybody. If you create awesome documentation of a product you you are enabling that product to the masses, because a great documentation of a project is uh my opinion, is the key of the success of that project.

C

B

Agree I like what jafar and johnny said so start contributing, no matter how documentation backfixes or code, please judge discussions as we say, please just start contributing join the community, that's a great uh uh wrap up and final for our show that uh um it's at the end. uh Unfortunately, we are at the at the end of our show. So uh I would like to thank everyone that joined it also in the chat and renato joined late, but renato you have the same.

B

If you use the same link, the recording will be available as soon we close the stream so use just the same link, and you can see the awesome demo from johnny johnny. Thank you very much for joining us. That was really great. Thank you. Thank.

C

You thank you so much for asking me. It's been a pleasure. It's been very funny and thank.

B

You yeah, we look forward to have you in the next episode jafar uh some appointment for openshift tv today, um yeah yeah.

A

Yeah so uh yeah thanks a lot jenny for the great demos. It was awesome. uh Everything worked, it was very well explained, so thank you very much uh yeah. So we still have some other shows coming up today. Of course, um if I may mention that, I'm I'm I'm going to be doing one this afternoon on pipelines as code, which is a very nice upcoming feature uh on openshift, which is basically how to allow you to push your tecton pipelines from the git application and that's another step towards get ups right.

A

So githubs we use to push like the resources and stuff like that. Now we can also push pipelines and then openshift figures out how to create the pipelines within within the cluster. So it's going to be today at 3, 00 pm uh and I'm gonna put the link.

B

Yeah, I put the link in the chat of the streaming calendar, so you can join that link and see all the episodes of today. But if you have a direct link to the show, please share in the chat, so the people can.

A

Join yeah, okay, yeah, so I just shared it. uh I believe on youtube: I'm not okay with stream chat, but if you yeah um natalie, if you can get the link from our zoom zoom chat and paste it through restream. That would be also helpful.

B

Yes, sir, uh let me let me copy from our chat, so we can share here. So this is the link to the um openshift tv uh show, which is will be on pipelines as a code right.

A

Exactly yeah yeah, so this is not uh exactly uh openshift pipelines like the the techtone downstream. It's a new feature that allows you to define your your takedown pipeline in your git repository and as soon as you commit your app changes.

A

Openshift triggers the pipelines automatically uh on openshift pipelines, it's similar to uh what we used to do with jenkins file, where you have a jenkins file in the repo, but it's more advanced because you can now react with pull requests and stuff like that. So stay tuned and.

B

This afternoon,.

A

uh I'm gonna be doing a demo, so I hope uh yeah. It's an interesting topic.

B

Can't miss it, it's of course it's another live demo, here's openshift tv, so I would like to close up and wrap up everything. Thank you. Everyone for joining the recording will be at the same link I saw while johnny were playing some telegram chat from our okd group. If you would like to join our okd channel, which is the openshift community group on telegram, please join by this link, I'm putting here in the chat and also, if you like, uh to follow johnny on twitter, his his twitter handle.

B

You feel free to to follow him and say something about this awesome demo. Today, uh java appointment today, openshift tv, the openshift tv coffee break, will come back on december, the 15th. We will have another up, dev topic uh with alex, groom and other folks joining us. So please uh stay tuned and see you on december, the 15th same time here at openshifttv.

B

Thank you, everyone for joining and see you next time. Ciao. Thank.

A

You very much thank you, johnny bye-bye.

C

Bye-Bye thank you.

A