Cloud Native Computing Foundation KubeCon + CloudNativeCon North America 2020 - Virtual, 4 Dec 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Sponsored Session: AWS - Say goodbye to YAML engineering with the CDK for Kubernetes

Description

Don’t miss out! Join us at our upcoming event: KubeCon + CloudNativeCon Europe 2021 Virtual from May 4–7, 2021. Learn more at https://kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects.

Sponsored Session: AWS - Say goodbye to YAML engineering with the CDK for Kubernetes

The CDK for Kubernetes (cdk8s) is a new open-source software development framework for defining Kubernetes applications and resources using familiar programming languages. In this session, we will show you how to define your first cdk8s application, define reusable components called “constructs” and generally say goodbye (and thank you very much) to writing in YAML. We’ll also introduce our construct library cdk8s+ and show you how you can accelerate application development and adoption using cdk8s for any Kubernetes cluster.

https://sched.co/fezY

A

Hi everyone glad you could join us. My name is ellie, I'm an engineer on the city cage team, and today I want to show you what it's like to develop: a kubernetes application using cdks and cdks, plus all right. So just some context. This is the construct catalog. It's essentially a twitter account that tweets every time a new cdk library is published.

A

Now today we want to add some search capabilities to it. We want users to be able to query to discover those libraries okay. So this is my application. It was generated or initialized using the cdk cli and you can see there's a main.ts file here, which declares my cdks chart. This is where my resources will go: okay to implement the the application we're going to use elasticsearch as the backend and, specifically we're going to use the elasticsearch operator custom resource. So to that end, I already have the custom resource definition in my kubernetes cluster.

A

You can see it here and what I want to do is import this definition into my application, so I can use it in my typescript code now to do that. I'm going to use the syndicates import command, which will read the definition, the schema the json schema and generate typescript code, all right. So let's do it so we run cube, cuddle, get crd and we take the elasticsearch crd output, it as a json and pipe it to cdk's import.

A

Okay. Now we have this elasticsearch.ts file here you can see it's pretty big and it contains the entire api spec of the custom resource as typescript code, and you can also see that there's this top level. Elasticsearch api object here or a resource, and this is what we'll be using all right. So, let's use it we're going to do a new elastic elasticsearch.

A

You can already see the nice id completion. I give it a scope and an id and some properties so we're going to give it you can see. There are a few optional and a few required ones. You can also see a nice inline documentation.

A

So, let's start with the version we're going to use 7.7.1, uh we'll configure onenote set with a count of one and the name of default, and we also need a few other configuration properties which I'm just going to paste in. So I don't make any mistakes.

A

Okay, I'm also going to disable ssl for.

B

A

Just because it makes it easier with the demo, so this is how you do it, and also I'm going to explicitly set my service ports, so I'm going to use the standard, oops standard, elastic, search, 9200, port, okay, that's it! That's all I need now. uh I can generate my manifest sorry.

A

I do this by running cydicates syn and let's take a quick look. So this is the manifest. uh No big surprises here right. This is what we expected, and also it has this metadata name here which was auto generated by the cdks.

A

So the city case will generate a name for every resource that doesn't have a name configured, and this is going to be very useful. uh We'll see why, later on, all right, let's apply this to the cluster.

A

Okay, let's take a look at our pods. We have one initializing pod. That makes sense. Let's give it a second now, let's take a look at our service, we have this http service, which we configured with the 9200 port, and you can also see that the service name also has this uh prefix of the custom resource name this one.

A

Okay, um let's see where our port pods are have one running pod: it's still not ready. Let's give it a second or two all right cool. It's ready! Now, I'm also going to index some mock data into the cluster. Just so we'll have something to play with. So I can do this by running cube, cuddle port forward, and I have this chrome command ready here all right now, let's, uh let's move on so the next thing I want to do is create the a small proxy service.

A

That's basically going to translate user requests into elasticsearch queries and return the result to the to the user. So I already have this code prepared. It's not super interesting. uh The one thing to note here is that I need to pass these three environment variables to it, and this is uh we're. Gonna have to pass it from the from uh from the application from the manifest definitions all right. So what I want to do now is create the container right.

A

I need to create a container and configure the environment variables and, uh and somehow include this query.js file in the container.

A

So to do that, I'm going to use cdksplus now citycaseplus is a package which is part of the citycade's toolchain, and it contains these high-level intent-based apis that are based on the low-level core kubernetes objects, but they make it a lot more easier and natural to work with. Let me just let me just show you what I mean all right, so I'm just going to import this cdksplus library as k plus from cdksplus, and I'm going to create my container.

A

So I'm going to do new k plus container, so I need to give it the image. So it's going to be node 12.18.0.

B

A

Stretch specific um and the command is going to be node, dot, js right and I'm going to give it a port, and I need to configure the port here as well. So let's just extract this out.

A

uh It's going to be query port and we're gonna use it.

B

A

And we're going to use it here. Okay, now I need to um expose or make this query.ts file available to the container. So I like doing this with config map based volumes and then just mounting the volume onto the container at a specific path. So let's do that? Let's create a config map, so I'm going to do.

A

Config map, k, plus config map, give it a scope and an id and take a look at its api. So I see that there's this add file api, which is exactly what I need.

A

So I'm going to use it, and now I need to create the volume right, so I'm going to do k plus dot volume from config map, and now I need to mount this volume to the container so to do that, I'm going to store the container again in constant and take a look at its api. So I can see that there is a container.mount api which accepts a path and a volume, and I'm going to need to configure this path, also as the working directory of oops of my container, so that this command will work.

A

Now, let's extract this out.

A

Okay and use it here and use it here, duplication averted all right now. I need to configure the environment variables so to do that, I'm going to use the end key and let's take a look at which environment variables do I need. So I need the elasticsearch username. That's easy!

A

um It's going to be like the standard default username that the custom resource creates, so I'm going to do k plus dot n value from value, basically just a way to pass a literal string.

A

The next thing is the elastic search endpoint, which is a little more interesting. It's going to be k, plus dot n value, but let's stop here for a second, so the endpoint is actually related to the service that we saw before. So, let's take a look at it again.

A

This is the http service that I want to hit, so I'm going to have to construct the endpoint using the service name and the service port with with the http of course. So again I'm going to do from value. So it's a literal string, it's just a dynamic one, and here I'm going to do http and now I need to have access to this uh generated name.

A

So to do that, I'm going to store a custom resource oops in a constant elastic, and now I can just use this resource.name and to act to get the name and again, this dot name will work for any uh cdks resource, even if you don't specify a name on your on on your own, all right so now I just need to append this suffix.

A

And I also need to give it the 9200 port, and since this is repeating itself also, so let's just extract this. So it's going to be es port and I'm going to use it here and I'm going to use it here.

A

All right next up is the password, so the password is the most interesting one because it's actually stored inside the kubernetes secret and it's created by the custom resource. So if we take a look at the secrets now, this is the secret we need and again you can see that it's prefixed with this resource name, which we already know how to access. So the first thing we need to do is import this secret into our manifest, so to speak, so to do that, we just use the k plus dot secret from secret name and again.

A

We already know how to create this name. Just by doing this.

A

Okay and now we can use the secret to create environment variable values, so we do k plus and value from secret whoops yep nope from secret. We give it a secret and uh we give it the specific key inside the secret that actually contains the value. And again this is the the documented uh key.

A

Okay, I think we're ready. As far as the container is concerned, the next thing we want to do is just create the deployment, so we do new k plus deployment.

A

We give it a scope and an id I'm going to go a little faster here um and configure so just one replica and the pod spec is going to be just my single container, oops container, all right and now I want to expose this deployment as a service right. So let's take a look at the api that the deployment resource has um okay employment dot. So you can see that there's an expose here uh which says, expose the deployment via service equivalent to running cube, ctl exposed. This is perfect.

A

This is what I need, so I'm just gonna give it the port just pour it uh to my uh willing um all right. I think that's that's that's that should work, so, let's generate the manifest and see and see where we're at so city, kate's synth.

A

Let's take a look all right, so you can see the manifest is much bigger now, obviously, also because of this inline application code here, but we can also see that there is the service here right and the deployment here, which is great and also we can see that these selectors are here, even though we didn't really specify them in the application right here, we didn't have to think about selectors like anywhere normally when you define the ammo, you have to uh apply labels to the pod, and then you have to apply the selectors or use them as the selector for the deployment and the same thing for the service.

A

Now cdksplus actually does that for us it it, it interprets our intent. This is what I mean by like intent based apis. It is. It assumes that if the pod spec template is defined in the scope of the deployment here, then it makes sense that the deployment will just automatically select the pods in its template. Now. Another thing worth mentioning is that, for example, this target port of the service is also kind of implicitly inferred right.

A

We didn't have to specify this here when we exposed the deployment, because the the information is already here and also uh citycatsplus configured uh this volume in the podspec. For us, we didn't have to configure uh the volume here in the pod spec, because again this volume is is attached to this container. So it knows that it needs to be part of the pod spec as well.

A

All right, um I think we're just about ready to apply this. So let's apply. Let's just do a one, quick, uh quick check to see that uh everything is configured properly, yeah all right, so the cube cuddle applies.

A

All right, let's take a look at our pods yeah. We have one running pod. So now comes the moment of truth: let's create the port forward and try to hit that endpoint. So we do cube cuddle port forward using the service name and the 9000 port that we configured here and let's hit the end point now.

A

Yeah cool: you can see that this is the cydicates cdk library that we saw before in the in the catalog this one the statics website- and this is what we're, what we're, after obviously uh building some nice uis on top of that all right. So this is uh what I wanted to show you. um I hope you can see the potential these kinds of apis have and how they can really reduce the cognitive load for developers and make it a much more friendlier experience.

A

We think it's it's a cool direction for for manifest authors, um all right, I'm gonna head over to nate right now for a quick recap and uh thanks so much bye. Everyone.

B

Thanks ellie, that was a fantastic demo. I really want to thank you for taking the time with us today to learn a little bit more about how cdks can help you accelerate and standardize development of kubernetes anywhere. We have big things planned for this project, so get involved or follow along on github or at cdks.io.

B

If you want to check out and and try the demo that ellie just showed, we have a link right here and that's on our github.