Kong Kong's Kubernetes Ingress Controller, 21 Sep 2021

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: APM with Prometheus and Grafana on Kubernetes Ingress

Description

In this video, we’ll see how operating a Kubernetes environment using Kong Ingress Controller, Prometheus and Grafana can simplify the daunting task of getting full application performance monitoring up and running. Try Kong Ingress Controller: https://bit.ly/3guN9D6

▬▬▬▬▬▬ TIMECODES ▬▬▬▬▬▬

0:00 Intro
0:26 Getting Going With a Kubernetes Cluster on kind
0:59 Installing Prometheus and Grafana
3:30 Installing and Configuring Kong Gateway
5:43 Final Server Configuration
7:03 Let’s Demo This Dashboard!
14:00 Conclusion

Follow along in the blog version: https://bit.ly/3tUHuN0

▬▬▬▬▬▬ COMMUNITY ▬▬▬▬▬▬

⭐ https://github.com/Kong/kong
🐦 https://twitter.com/thekonginc
🌎 https://konghq.com/community/
🍻 https://www.meetup.com/pro/kong
❓ ️https://discuss.konghq.com/
💯 https://konghq.com/kong-champions/

#Prometheus #Grafana #Kubernetes #APM #Microservices

A

While monitoring is an important part of any robust application deployment, it can seem overwhelming to get a full application performance monitoring stack deployed today. We're going to take a look at how we can leverage the power of kubernetes prometheus grafana and kong to simplify this seemingly daunting task.

A

We're also going to be making use of kind, kubernetes and docker to give you a local kubernetes installation right on your development machine head on over to your terminal and make a new cluster with kind if you've got it installed. This might take a few minutes so we'll check back in a second.

A

Once kind has done its thing and giving you a local cluster, you're gonna need to get a couple. Repos added for some helm, charts first one's gonna be for prometheus, related, helm, charts and we'll have some default values set up as well.

A

Really all we're doing here is setting a scrape interval for prometheus to check whatever data sources. It's going to be going out to every 10 seconds.

A

Use the helm chart to get prometheus installed.

A

We're making a new namespace called monitoring here to make sure that everything stays sealed in its own little box.

A

And then we're going to do the same thing for grafana.

A

Using the official grafana helm, charts.

A

And grifana's setup is a little bit more involved.

A

Make sure you use some persistent volumes, we're going to be setting up a default data source that'll get in touch with the prometheus server.

A

We're also going to be telling grafana that we want to use the official kong dashboard.

A

So, let's then get that installed.

A

Let's see here, it gives us some instructions about how to get our password for our admin account we'll make use of those in a little bit.

A

Now, last but not least, we want to install our repo for calm.

A

Again, we're making a new namespace just for our kong instance, keep everything nice and tidy.

A

Now we need to make sure that everything fits together.

A

Just so you're aware we are using these values on our kong pod, make sure that it goes out and talks to prometheus properly and.

A

This plugin is going to make sure that kong gets hooked up to prometheus. So let's get that applied.

A

Now, at this point, we should have everything wired up together and can move over to a new terminal window and start setting up some port forwarding rules so that we can take a look at what we've made in our browser.

A

First, we're going to get the pod name of our prometheus server.

A

Set up a port forwarding rule over to port 9090.

A

We'll do the same thing for our grafana server.

A

This time we'll use port 3000.

A

And finally, we'll get the kong.

A

A

And set up the port forward to port 8 000.

A

And if you've done everything right, you should be able to go. Take a look at these in your browser. You can look at localhost, 9090 that'll get your prometheus window, you can look at localhost, 3000 that'll, give you your grafana window and if you look at localhost 8000 it doesn't really show you much, but this is in fact our kong instance.

A

Now we're going to want to get into our grafana admin panel.

A

Obviously we don't have a password, except for the one that's been automatically generated, so we'll use the command that was given to us as we installed our grafana instance, and this password isn't going to be anything special. It's just generated on every installation.

A

You can see that I grabbed this from the terminal with that uh special get secret command and now that we're logged in we can scroll down and take a look at this official kong dashboard.

A

This is the real test. If you can see these gauges with numbers that aren't just no data, you have done something right.

A

I see it's been a couple minutes actually in real time and there's some things going on already over here and you can take a look around, but really once we start opening up these bigger sections, there's no other data. So let's get some other data going.

A

If we come back over to the terminal, I've gotten some services mocked up, that'll use, http bin, which will simulate a bunch of different room response codes, and you can see that I've got billing. I've got invoices and I've got just a basic http server here.

A

So this sets up billing invoices and comment endpoints that we can put on our kong instance, and normally this is where you deploy your actual services to that endpoint and, and you would then be able to operate your system pretty seamlessly that way. So, let's apply that.

A

Okay, you see it's going to be creating all three of those services. It's giving us a http server to to really serve all of those things, and we also need to look at our ingress.

A

Is going to be using kong and it gives us the path for each of those services that we've just decided to build. Let's apply that as well.

A

Now, to get everything really going, it takes about a minute or two, and uh once you have all of your services going, you can set up this lovely while loop, that's going to simulate some 200 uh responses, some 501, some 201s and some 404s. This is really just going to give us dummy data in the real world. You could deploy this apm stack behind your real production services and you'd start getting some information right away, as you can see, we're just getting a steady stream of requests to 8000.

A

So we'll let this run for a little while.

A

I've let that while loop run for a couple minutes and now I've come back to my computer and when I refresh the data, I've got a little bit of a different look here. This line has jumped up pretty high or about 24 requests per second, but we don't see very much, but that's because we're looking at the last 24 hours. So if we switch the last five minutes, we can see that we've already got some pretty decent data here.

A

Scrolling down, we've got our requests per second broken down by route and service. We've even got the return codes from each of those requests and there's a whole bunch of them, but, as you can see right out of the box lots to work with.

A

If you look at the details here, there's even latency for all sorts of different pieces and really just an incredible batch of information right out of the box here from this kong official dashboard in grafana.

A

Some areas don't have anything some do, but all told really impressive stuff.

A

The really beautiful thing about all of this is this is built on top of prometheus.

A

So if you want to take a look at building your own metrics, you can, if you want, to go and build another dashboard, that's another option as well all told we've got a really amazing solution without much serious effort on our part, just a few helm, charts and configuration files and we've got a full stack. So, as you can see, this is uh this is a lot more possible than it might have felt before.

A

There's really no limit to what you can do with this stack.

A

Now, since we are doing this locally on our own computer, we are going to want to close down our while loop and we're going to want to come over here and bring these port forwards to a close as well.

A

That's all of them, so we can close out of there and then really the really amazing thing here. If you're all done, you can come in here and delete the whole cluster and there's no no after effects on your on your local environment.

A

Thanks so much for watching- and hopefully you learned something today and are able to take this back to your own systems and and build something amazing with kong and all these other great tools thanks so much.