keptn Keptn Lifecycle Toolkit, 30 Dec 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Keptn Lifecycle Toolkit Demo Tutorial on k3s, with ArgoCD for GitOps, OTel, Prometheus and Grafana

Description

Keptn Lifecycle Toolkit (KLT) gives observability and orchestration of your GitOps deployments. It does this by ...
... using OpenTelemetry to trace any deployment on K8s from commit to running
... providing Serverless functions to execute pre- and post-deployment checks
... pulling Prometheus to validate environment health (SLOs) pre and post deployment
... bringing Application awareness to your Deployments

In this new tutorial - https://github.com/keptn-sandbox/klt-on-k3s-with-argocd - we walk you through through a fully ArgoCD, OpenTelemetry, Grafana and Keptn setup in 5 minutes on a bare linux VM using k3s as a lightweight K8s cluster.

Watch this video to see what you get after 5 minutes. Learn how Keptn Lifecycle Toolkit works and how it seamlessly provides observabilty and orchestration of your GitOps implementation

To learn more check out: https://lifecycle.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

Hi there I spent some time to look into the captain life cycle. Toolkits, there's some great demos from Thomas shoots already out there, but I thought I want to create my own demo, based on k3s lightweight kubernetes, using Argo CD as the github's tool, using Prometheus using grafana and also slick. So what I've built now is a demo that you can just run on any Linux machine, in my case, I'm using AWS ec2 instances, and it takes about really five minutes and what it does. It sets up.

A

Argo kept lifecycle toolkit, obviously installs Prometheus grafana exposes everything through in Ingress and then also uses a sample app and it integrated with this with slack meaning. Every time you deploy with Argo Captain will intercept the deployment and will then execute pre and post deployment tasks, as we call them and I just chose to use a select notification. So there's multiple options you have, you can either do the manual install. So you learn every step or you just go through the automated install where you just need to clone that repo.

A

You need to Fork it as well, because I have the sample app and you obviously want to then deploy different versions of your app. So you can make these changes in your git repository optionally. You can also install Dynamics with one agent. If you expose these variables, you can also do the select notification, and then you just execute that script install klt on k3s, that's it and at the end you get the output. So what I've done? I have a forked, this repo, because then I can always make changes to my sample app.

A

As you see here right. This is the simple node Def, and basically you see here the captain tasks for sending the select notification, there's also Captain tasks that I've prepared for evaluations. This is for Prometheus querying slos. The real thing, though, is, and the only thing with that you need to do with any type of app.

A

If you have your deployment, the only thing you need to do- and let me zoom in here is uh you need to annotate your deployment with uh Captain workload, cap conversion, and then you can specify their pre and post deployment tasks or evaluations and I'm just referencing here the task so I've I've installed all this. Where did I install it I have it here on my ec2 machine. That means really what I've done.

A

I've just cloned the repo I checked it out and then I ran install klt, and if you look at this get parts then right. This was all it was installed, there's obviously the core. There is all the monitoring components. It's Prometheus, there's Argo that is installed uh whether it's the lifecycle toolkit itself is installed and I've already run a couple of deployments. This is why you see the simple node div namespace, with my sample app. So now let me go.

A

Let me go back really what you have uh the the app will leverage uh your IP address, the public IP address of your machine I'm using nip. I o here. If you have your own DNS, then obviously you can use your own DNS, but by default I'm, just using the public IP address, so that means I can go to Argo CT europe.nip.io and after about five minutes of running the script, you get to see Argo pre-configured with the app already, because the install script also installs the app.

A

What else do we have grafana grafana dot same thing right? You get full access to the dashboards, there's a couple of dashboards here that come with the install. If you go in here, let me zoom in a little bit the default dashboards, Captain applications, lifecycle, toolkit, metrics workloads and so on. You can see here a lot of cool uh default dashboards that we have. Let me go back to the captain applications you can see here. Over the last three hours, I ran a couple of deployments, they're all successful.

A

uh The team did a great job and also providing metrics around time between deployments deployment time itself. How many are successful? How many have failed your Dora metrics and, if I scroll down here, you also see the traces, because one of the things that they kept in lifecycle toolkit does pretty well in the beginning of like kind of the core capabilities, it's automatically giving you an open telemetric distributed Trace. So when I deploy a new version of my app I automatically get an Hotel Trace with that includes everything that they kept lifecycle.

A

A toolkit does uh pre-deployment validations the actual app deployment when kubernetes deploys the app and also any type of post deployment notifications. Now the only thing I need to do in case I want to. You know basically deploy a new app. So this is now my clone repo. That's basically the repo that Argo right. Here's Argo Argo has this repo here in the Argo app, that's my klt demo.

A

The only thing I want to do is if I want to change the deployment I'm just going in here and make a quick change. I'm going from version 30 to version 4.0, so I'm changing the captain version information here and I'm also changing, obviously the container itself and then I'm just doing updates to V 401 and I commit what that means. Is red. Argo will now hopefully pick up that change. Let's see refresh here, we go it's progressing.

A

um It will hopefully sync soon right, because Argo is my github's operator. It will deploy my changes, but then Captain life cycle toolkit really hooks in, and uh here we go, we see simple notice being deployed. The captain lifecycle toolkit kicks in does pre and post deployment checks and, in my pre-deployment check, I have specified. If you remember on my repository. Let me go back as a pre-deployment task. I want to execute the notify test and notify task is a JavaScript function here on the captain tasks right, I just use this one here.

A

This is actually from the example that was put out there by Thomas. Captain allows you to Define. Any type of JavaScript will be more features coming, but what I'm basically doing here, I'm just sending a select notification, so everything happens through configuration by the way. Here is my app. That was the one before I think now build number three that was the old one that is deployed now, version 4 will be deployed um and, as this is happening, let's keep an eye on it.

A

I should, if I refresh here, see my deployments going through here, we go simple. Node right still, deploying uh I should see in my dashboards if I refresh here, let's see if I get all the numbers deployments, not last three hours, you know that's six hours, let's go back a little bit been doing stuff over the time. Simple note should see an active deployment coming in hopefully anytime soon. Actually, this is probably the active deployment right now. That's the 1003, that's also the distributed Trace, so I get a trace.

A

That shows me what's happening right now. I see that the lifecycle toolkit came in for version 401. It's doing it already did the pre-notifications. Now it's actually doing the uh the workflow deployment itself, as the pre-notifications have already been executed. I should be able to go over here here we go 1104. That's my local time. I got two notifications because I specified a pre and post, which means by now I assume the deployment is already done. That means this app should now be here.

A

We go version number four um I should see here all of the information about time between deployments deployment time uh if I go even further right, because yesterday I did some tests here, I think so at least actually last two days probably had a couple of more uh data points here, but yeah as you can see here, you get automated visibility. uh You get automated distributed tracing and everything is easily configurable uh through cfds, because this is what the lifecycle toolkit is really all about right.

A

You only have to specify your captain configuration tasks and evaluations as cfds, and then you need to annotate your deployments, like I did here, Captain workload, Captain version the pre and post deployment stuff. This basically tells Captain what actions to execute and yeah.

A

One thing that I've forgot to mention in my example: the namespace itself needs to be annotated that you want to enable the captain lifecycle toolkit on that namespace, because technically what Captain does it extends your kubernetes scheduler, and we only do this on those deployments that are in the namespace that is enabled for the lifecycle toolkit so to get started.

A

um If you want to run this demo, like I, did let me scroll up again: Captain sandbox klt on k3s, with Argo CD and all the rest. I hope this will help you just as this helps me takes five minutes to stand up the whole demo and then play around with it um yeah. Thank you. So much and uh I would say: happy New, Year cheers.