keptn Keptn Maintainer Track, 4 Oct 2022

Previous Meeting

⏯

youtube image

►

From YouTube: Keptn Lifecycle Controller v0.1.0 Demo - October 2022

Description

Learn more: https://keptn.sh
Get started with tutorials: https://tutorials.keptn.sh
Join us in Slack: https://slack.keptn.sh
Star us on Github: https://github.com/keptn/keptn
Follow us on Twitter: https://twitter.com/keptnProject
Sign up to our newsletter: https://bit.ly/KeptnNews

A

Hello, everyone today I'm very excited to give you the first demonstration of the captain lifecycle controller. The kept lifecycle controller makes it easy to manage the life cycle of cloud native apps without writing operator code. We all know that operators are very nice concept and work perfectly fine for commercial aftershave software, but they are very hard to adopt when the life cycle of an application changes and very hard to write.

A

Initially, the lifecycle controller makes it easier to attach task handles to pre to print post deployment events and should also be able to attach into kubernetes events in the future.

A

This could be in form of typescript functions, but also this could be containers in CLI in and Cloud events in the future. In today's demonstration we will focus on typescript functions.

A

um The captain livestock controller will make the apply effective, better Observer, with traces and Matrix in the future. So what we will see here or today is that we can do all of this, so pre-post deployment tasks um and the deployment of an application doing a simple Cube cuddle, applying minus F Dot in our pre-deployment task. Today, we will check if one of the workloads is ready, so we will deploy a simple Potato Head application, which has around six Services.

A

One of them is the entry service and all of the other services or workloads will check if this service is available before they start deployed after the deployment, we will run one post deployment task, which is one simple, select notification in my case, so um I think I've talked enough. Let's go to the demonstration, so um the LA, the captain lifecycle controller, can be installed very easily with two cubesatel commands. The first one is for installing this web manager, which is a dependency at the moment. The second one is the lifecycle controller itself.

A

What we'll deploy today is the Potato Head, and we will deploy this using a simple kubernetes manifest.

A

um So what we'll see in this manifest is that we have a namespace, a deployment as we know it, and the First new thing we see here is that we annotated the workloads with some Gap notations.

A

The first one is the app which specifies to which application this belongs to the second one is the workload name. So in my case, this is the potato head entry here. The third one is the version of this application of this workload. Sorry and the fourth one is the post deployment task which will be triggered. So this is called post deployment, Halo Empire case here for the entry service. We also added the init container to see that that we can wait until this is deployed with our post deployments uh checks.

A

Then we have another few services, such as the Potato Head head service, which is also annotated with another workout name um same version name, but it also has some pre-deployment task, and this is called check. Entry, so is in this case, and this will check if the entry service is reachable before it gets before. The head service gets deployed foreign, so we talked about tasks here and they have to be specified in some form.

A

So, when I take a look on the check entry armor here as here, the typical Captain task definition and the name I am using here so check. Entry service is exactly the same as I used in the annotations before, in this case, I specify that this task definition uses a function which is referring to an HTTP source, and this is um located in my in our lifecycle controller repository and contains the function for http.

A

Furthermore, we pass over one parameter to this to this function, which is the URL we want to check, and if we take a closer look on this function, we see that this is a simple task: a typescript function in Arduino runtime, which gets data from an environment variable which is called beta. So this is the thing we did here in our parameter.

A

We specified here in our parameters and in the end, we simply try to fetch the URL we got here and until we- um and until this is not finished successfully, we will also we will always exit with an error which will cause the port to restart okay, so these um uh and for the Post deployment task.

A

We do a simple select notification and one thing which is different to the pre-deployment task before is that we specify our function inline in this case, so you can also add your typescript code in in the custom resource definition and the captain. Lifecycle controller will also um execute this one.

A

Okay, so I think we have everything we need, so we can try to find and try to um execute all of this.

A

So what you see in my shell screen here is on the right upper side, the ports which are which get created. So at the moment the namespace is empty and in the lower um part of the of the window, we see the captain workload instances. So the version of one workload which are currently deployed and in which state they are and in the directory I am at the moment there are all of the files um you saw in the repository before one file is new here.

A

This is this like secret file and this it specifies where on the web Hook and the things which should be sent to this like Channel afterwards, so let's simply apply this.

A

And with my teeth as a lot is happening here, so on the right upper side, we see that um one container is initializing. This is the entry Service as we specified, and this will wait for around 30 seconds. All of the other things are pending, because this is services, not services, not running at the moment. What we see on the upper part of this is uh these are the pre-deployment tricks which are currently erroring, because the service is not started at the moment and not reachable.

A

So after the 30 seconds and after the unit container has finished, we see that this is in a running state, which is also reflected in the record instance state. So we see that the pre-deployment has succeeded. The deployment itself also has succeeded, and the poster brand is also finished and in the right lower side of the of the screen. You should see my slack window where you see that I got a notification that the entry service is deployed.

A

Now that the service is reachable, all of our pre-deployment checks could be completed, and what we see now is that all of the other ports are also starting. Now and now part workable workload. The other service will also succeed and in the end, as we have no post deployment tasks for these Services, everything is running perfectly fine.

A

So, let's get back to the presentation.

A

So what you have seen in this short demonstration was that we can use shared functions for your pre-deployment checks. For instance, you can take all of you and you can have all of your pre-deployment tasks in one repository on one server and simply reference to them.

A

You can also use inline functions, so you can specify your functions in a custom resource definition and use them for your pre or post deployment tasks.

A

What we also saw very well is that the lifecycle controller holds back the deployment until the pre-deployment tasks have finished and last but not least, everything can be done using cubes details, so we did not run any pipeline tooling and we did not run any external tooling. um We all did. We did everything in a cloud native way applied only manifests and in the end our deployment got finished very well.

A

So um if you liked this demonstration- and you think that you want to try it and try this out and I hope you will um everything you saw here- is in the repository of the captain lifecycle controller, and if you want to participate in this project and in everything we do around application lifecycle. There is a working group around this, so on the app lifecycle working group after you try it all of this.

A

um It would be very nice if you would share your thoughts and you could pin me always on the captain's leg or also participate in the app lifecycle working group channel on the captain, live on the captain's leg and last but not least, um let's make it better. So if you find some issues or if you think that some some things could be, um could be solved better, um just open up an issue, um I think you can also.

A

You can also do code contributions and if you would like to see some more features and so on philosophy, read opinion on the captain's leg and let's discuss these things so I hope you had as much fun as me and I hope. You have a lot of fun with the captain lifecycle controller and have a nice day and see you bye.