Container Orchestration with Kubernetes, Part 5

Managing Kubernetes

Hey everybody! In this post, I’d like to talk about Container-centric infrastructure with Kubernetes. This is Part 5 of a series of posts on using installing and using Kubernetes.

If you missed Part 4 where I give you Install Kubernetes on your Computer (Minikube), make sure to check that out.

In this post I’ll show you two ways to manage your Kubernetes applications:

  • kubectl
  • Kubernetes Web UI Dashboard

The kubectl command

When Kubernetes starts a pod it’s isolated in its own network. So how do we interact with the application?

One way is through a proxy, which you can start in a separate terminal window. Start a new Terminal window and execute the kubectl proxy command.

kubectl proxy

By default the proxy uses port 8001, but you can specify a different port using the --port option.

kubectl proxy --port=8080

This starts the proxy using port 8080. Now you can connect to the cluster through port 8080 on your computer.

I’ve written a script to get the pod name and use it to access the application through the cluster’s REST interface. The script is called test-k8sdemo.sh and takes the port as an argument.

./test-k8sdemo.sh 8080

And there you see the output.

Now that we have an app up and running, let’s talk about how to manage it. You can use the kubectl command to get information about pods, deployments, and so on.

For example, you can show a listing of the deployments:

kubectl get deployments

Or of the pods:

kubectl get pods

You can even get detailed information about the pods:

kubectl describe pods

This produces a lot of output, but it’s valuable (if somewhat verbose) information.

Kubernetes Web UI Dashboard

Now I love the command line as much as anybody, but did you know that Kubernetes has a Web-based UI dashboard?

By default the Dashboard is not deployed. From the documentation, you can copy the command to create the deployment through the kubectl command:

kubectl create -f https://raw.githubusercontent.com/kubernetes/dashboard/master/src/deploy/recommended/kubernetes-dashboard.yaml

Paste the above command into the console to deploy it.

Make sure the proxy is running (I showed you how to do this in Part 4). Start the proxy in a separate Terminal window:

kubectl proxy --port=8001

Once the Dashboard is deployed you can access it from http://localhost:8001/ui.

You can see the same basic information as from the various kubectl commands, but in a UI. Here’s an example of what I see using Chrome:

Conclusion

Well, that’s it for this series. I hope you enjoyed it, and make sure to check out the video below that walks you through everything I covered in this blog series (it’s true what they say: a video is worth a thousand pictures!).

Want to know more?

If you want to learn more about Kubernetes, check out this post at IBM developerWorks: What is Kubernetes?

Be sure to check out the other posts in this series:

Check out this video at IBM developerWorks TV, where I show you everything I covered in Parts 1-5 of this series:

Introduction to Kubernetes:

Thanks for reading!

–jsp

Container Orchestration with Kubernetes, Part 4

Working with Kubernetes on your Computer (Minikube)

Hey everybody! In this post, I’d like to talk about Container-centric infrastructure with Kubernetes. This is Part 4 of a series of posts on using installing and using Kubernetes.

If you missed Part 3 where I give you Install Kubernetes on your Computer (Minikube), make sure to check that out.

The Application

So now you’re finally ready to start using Kubernetes. The first step in deploying a containerized application is, well, the application. I’ve put together some demo JavaScript code that you’ll run as a Node.js application.

Open a Terminal window or command prompt, navigate to a location on your computer where you want the code to land and clone the code from GitHub.

git clone https://github.com/makotogo/developerWorks

Navigate to the kubernetes directory and take a look at the JavaScript code. This toy application doesn’t do anything serious, just outputs a message to the console. But it’s enough for our purposes here.

Now run the application in Node.

node k8sdemo.js

Open a browser and point it to http://localhost:8080 to see the message.

And there it is. Now that you have an application, it’s time to containerize it.

Containerize the application

Here’s how you use Docker to do that. Take a look at the Dockerfile:

FROM node:6.11.4
EXPOSE 8080
COPY k8sdemo.js .
CMD node k8sdemo.js

Use this Docker definition to containerize the application. But before we can do that we need to start minikube, and I’ll show you why in a minute.

Execute the minikube start command:

minikube start

The first time Minikube runs, it has to download the ISO image to run inside of VirtualBox, and may take a minute or two depending on the speed of your internet connection. Once the image has been downloaded, minikube will start.

Minikube runs inside VirtualBox in its own VM. We need to make sure that Docker uses the internal Docker registry on Minikube’s VM by executing this command:

eval $(minikube docker-env)

Otherwise Docker will use the local registry on your machine, which is not accessible from the VM where minikube is running.

Now, from the Terminal window, execute the docker build command:

docker build -t k8sdemo:v1 .

This will create the image with a name of k8sdemo and a tag of v1, and store it in the Docker registry on the minikube host VM. When deploying the image to Minikube, you’ll reference that tag explicitly.

From the Terminal window execute the kubectl run command:

kubectl run k8sdemo --image=k8sdemo:v1 --port=8080

And now the application is up and running in its own Pod.

Conclusion

In this post I showed you how to work with Kubernetes on your computer. You containerized a simple application, then deployed it to your local Kubernetes cluster.

In Part 5, I’ll show you how to manage an application running in a Kubernetes cluster using kubectl and the Kubernetes Web UI Dashboard.

Want to know more?

If you want to learn more about Kubernetes, check out this post at IBM developerWorks: What is Kubernetes?

Be sure to check out the other posts in this series:

Check out this video at IBM developerWorks TV, where I show you everything I covered in Parts 1-5 of this series:

Introduction to Kubernetes:

Thanks for reading!

–jsp

Container Orchestration with Kubernetes, Part 3

Install Kubernetes on your Computer (Minikube)

Hey everybody! In this post, I’d like to talk about Container-centric infrastructure with Kubernetes. This is Part 3 of a series of posts on using installing and using Kubernetes.

If you missed Part 2 where I give you Kubernetes Overview, make sure to check that out.

In this post, I want to walk you through installing the software you need to run Kubernetes on your computer.

Install the software

You’ll need a few pieces of software to run Kubernetes on your computer:

  • Docker
  • VirtualBox (or some other virtualization software)
  • Node.js
  • Minikube

Install Docker

First, you need Docker. You can find instructions for your platform by clicking this link. If you already have Docker installed, skip to the next section.

Scroll down until you see your platform, which is MacOS in my case, so I’ll click on Docker for MacOS. I want the distribution from the Stable channel, so I click on the link that says Get Docker for Mac (Stable). Always verify the 256 bit SHA checksum of the downloaded file matches what is published.

Now double-click the DMG file to begin the installation, and follow the instructions.

Install VirtualBox

Minikube requires a virtualization environment in order to run. If you want to use software other than VirtualBox, check out the Getting Started with Minikube page at kubernetes.io. If you already have VirtualBox installed, skip to the next section.

To install VirtualBox, go to the VirtualBox installation page, and click on the link that matches your platform. ALWAYS verify the SHA256 checksum matches the downloaded file.

Now double-click the DMG file to begin the installation, and follow the instructions.

Install Node.js

In order to test the demo application, you’ll need Node.js installed. I already have Node.js installed on my Mac, so I’ll skip that for the video. Make sure to check out the Node.js installation page. If you already have Node.js installed, skip to the next section.

Click on the download link that matches your platform, and follow the instructions.

Install Minikube

Go to the Minikube page at GitHub, and look for Installation.

I like to use Homebrew to install software on my Mac. If you’re using Linux or Windows, the Minikube installation page has instructions for those platforms as well.

Drop out to a Terminal window and execute the brew cask install for Minikube:

brew cask install minikube

Homebrew installs Minikube and now it’s ready to go. Yes, it really is that easy (on MacOS anyway).

Conclusion

Now you have all the software you need to work with Kubernetes on your computer. Stay tuned for Part 4, where I’ll show you how to work with Minikube and define an application, deploy it to your local k8s cluster.

Want to know more?

If you want to learn more about Kubernetes, check out this post at IBM developerWorks: What is Kubernetes?

Be sure to check out other posts in this series:

Check out this video at IBM developerWorks TV, where I show you everything I covered in Parts 1-5 of this series:

Introduction to Kubernetes:

Thanks for reading!

–jsp

Container Orchestration with Kubernetes, Part 2

Kubernetes Overview

Hey everybody! In this post, I’d like to talk about Container-centric infrastructure with Kubernetes. This is Part 2 of a series of posts on using installing and using Kubernetes.

If you missed Part 1 where I talk about “Why Containers?”, make sure to check that out.

In this post, I want to give you an overiew of Kubernetes.

Pods

(Source: https://kubernetes.io/docs/tutorials/kubernetes-basics/explore-intro/)

A Pod is where an application is deployed, and contains resources specific to an application, such as file storage.

Pods can contain multiple applications, and all applications in a Pod share the same IP address and port space. Think of a Pod as the logical host, and is the atomic unit of deployment and scheduling in a Kubernetes cluster.

Nodes

(Source: https://kubernetes.io/docs/tutorials/kubernetes-basics/explore-intro/)

A Node is a physical or virtual machine in the cluster, and runs one or more Pods.

Each running Pod with the same definition is called a replica, and a Pod can be instantiated more than once to improve application scalability within the cluster.

Desired state of the cluster

The job of Kubernetes – specifically the master – is the maintain the desired state of the cluster at all times. But what is the desired state?

Let’s start with the documentation.

The desired state is defined as the applications and workloads you want to run, what container images to use, the number of Pod replicas, and other stuff like network and disk resources that your applications need. The master compares the current state of the cluster to the desired state and if they differ, makes any necessary adjustments until the desired state is once again reached.

This is helpful for, say, when an application crashes. The master will detect the application is no longer running, and restart the it. Now the cluster is back to the desired state.

Conclusion

In this post I introduced you to Kubernetes.

Stay tuned for Part 3, where I walk you through installing all of the software you need to run Kubernetes on your computer!

Want to know more?

If you want to learn more about Kubernetes, check out this post at IBM developerWorks: What is Kubernetes?

Be sure to check out other posts in this series:

Check out this video at IBM developerWorks TV, where I show you everything I covered in Parts 1-5 of this series:

Introduction to Kubernetes:

Thanks for reading!

–jsp

Container Orchestration with Kubernetes, Part 1

Why Containers?

Hey everybody! In this post, I’d like to talk about Container-centric infrastructure with Kubernetes. This is Part 1 of a series of posts on using installing and using Kubernetes.

Let’s start with the kubernetes documentation.

(Source: https://kubernetes.io/docs/concepts/overview/what-is-kubernetes/#why-containers)

Host-Centric infrastructure

Before containers, we installed applications directly on the target host, along with libraries our apps needed, plus the libraries any other applications needed as well.

At best, it presents DevOps with maintenance challenges. And at worst, integration conflicts between applications, versions of dependent libraries, and so on, can lead to poor application performance and even downtime.

Container-Centric infrastructure

With containers, on the other hand, all the applications run in their own isolated mini environments, away from each other, and the underlying OS.

And this is what you get with Kubernetes!

Conclusion

Container-centric infrastructure like that provided by Kubernetes has tremendous benefits, and in Part 2 I’ll give you an overview of Kubernetes, and talk about Pods, Nodes, and the “desired state” of the cluster. Stay tuned for that.

Want to know more?

If you want to learn more about Kubernetes, check out this post at IBM developerWorks: What is Kubernetes?

Check out this video at IBM developerWorks TV, where I show you everything I covered in Parts 1-5 of this series:

Introduction to Kubernetes:

Thanks for reading!

–jsp

What is Apache Kafka?

Change the way you think about data.

You’ve probably heard of Apache Kafka. You know, that Big Data, publish/subscribe message broker, time ordered log thingamajig that integrates all your application data, and provides real-time materialized views while being crazy scalable?

Yeah, that one. In this article at the IBM developerWorks Blog, I talk about Apache Kafka. I’ll assume you’re the  perfect candidate for applying Kafka, and talk about your architecture today. Then I’ll describe where you want to be, what is blocking the path to your getting there, and finally, how Kafka can help.

Check it out if you get a chance. Kafka is really cool.

And don’t forget to follow my blog, and follow me on Twitter @jstevenperry

–jsp