COMP4000 Implementation Experience 1: Getting Started with Kubernetes

In this implementation experience, you will be configuring and deploying a simple “hello world” Kubernetes cluster using minikube. While your cluster is not yet truly “distributed”, it behaves just as a real Kubernetes cluster would. For more information on Kubernetes configuration, you may wish to consult the official documentation. Links to specific documentation items will also be provided as hints later on in this document.

With the exception of Part 2, completing this experience shouldn’t take more than a couple of hours. Feel free to collaborate with other students if you get stuck. However, you must acknowledge any collaboration. Additionally, copying and pasting or simply “changing up” each other’s answers will be treated as an academic integrity violation.

Submissions

Your experience report should be submitted as a PDF file on Brightspace, written in paragraph form. Code snippets or screen shots are allowed to augment your prose but are not required. We will post a submission link on Brightspace within the next few days.

The experience report is due on November 2nd, 2021. However, it may be a good idea to get started as quickly as possible, especially if you plan on doing an implementation-focused final project.

Receiving Your Grade

This experience is broken up into three parts:

A series of easy tasks designed to get you familiar with setting up a simple Kubernetes cluster.
A harder challenge that involves you deploying your own image in your cluster to interact with a simple API.
An opportunity to reflect on your experiences with Kubernetes and make connections to overall themes in the course.

Students are expected to complete Part 1 and Part 3 to get a grade of at most a B+. Part 2 is optional, but must be completed to receive a grade of A- or higher. Marks will be deducted for insufficient explanations or answers that demonstrate a lack of effort. By this logic, you should have a fairly clear idea of what grade you will receive when you submit this experience, based on how much effort you put in.

Getting Started

To get started, you will need to run a few commands on your OpenStack course VM to set up the environment. If you are not yet on OpenStack, please get set up as quickly as possible. We have a detailed setup guide available if you need assistance.

SSH into your VM using your preferred SSH client. Remember that your username and password are both student by default.
If you haven’t already, change your password with the passwd command.
Set up a simple minikube cluster using minikube start. After running this command, you should see some information printed to your terminal as minikube sets up your environment. Let the command run to completion.
Download the deployment.yml file to your VM using wget https://homeostasis.scs.carleton.ca/~soma/distos/2021f/experiences/deployment.yml.
Deploy the experience 1 configuration by running kubectl apply -f deployment.yml. This should download some container images and create the necessary Kubernetes API objects to run our simple “distributed” application.

Part 1 (Easy)

Follow the instructions for each of the following numbered tasks. Make an effort to answer the accompanying questions, but more importantly please note down all of your observations and describe what you did for each task. You should also feel free to write down whatever questions you may have about a given task.

To achieve the best possible grade in this section, you must demonstrate that you have made an effort to understand the results of each task. (Note that an effort does not strictly mean a full understanding; it is okay to have questions!)

Kubernetes deploys Linux containers as “Pods”, which form an even higher-level abstraction around the container itself. In addition to Pods, Kubernetes supports several other API objects that abstract over various distributed systems concepts. The simple deployment you have installed uses a few of these API objects.

Using kubectl get <object type>, kubectl explain <object type>, and kubectl describe <object name>, identify at least three different kinds of API object that are used in our deployment and briefly describe what they do. You can refer to kubectl --help as needed to explain what these commands do.

Hint 1: Some resources are encapsulated by multiple layers of abstraction. Each “layer” counts as a unique type of object for the purposes of this question.

Hint 2: The command kubectl api-resources will enumerate a full list of all supported API objects. You may also wish to consult the relevant documentation if you are stuck.

Hint 3: You may want to have a look at the contents of the deployment.yml file.
Run kubectl get pods to see a list of the pods that are running in your cluster. You should notice several comp4000server pods and one comp4000client pod. Spawn an interactive shell into the comp4000client pod using kubectl exec comp4000client -it -- /bin/sh. Examine the layout of your container’s filesystem. What files and directories exist? Do you think these files exist on your VM or somewhere else? What commands can you run?

Hint: Try running cd .. to leave the empty /client directory.
Our cluster exposes the comp4000server pods using a NodePort service. To find its IP address, start by spawning another terminal by starting another SSH session. (You should probably have at least two terminals open for the rest of this experience.) In your second terminal, run kubectl get services and copy the CLUSTER-IP next to the NodePort service.

From inside your comp4000client container, use curl <IP address here> to send a GET request to the NodePort service. Repeat the same command a few different times. What do you notice about the output? Explain.

Hint: Try running kubectl get pods again to refresh your memory about the topology of our cluster.
The simple web server running behind our NodePort service exposes a convenient end point called /crashme that bricks the server. Try calling this end point several times like for i in $(seq 1 20); do curl <IP address here>/crashme; done, then try interacting with the servers after a few moments.

Repeat the same experiment, but this time run kubectl get pods -w in another terminal to watch the status of your pods in real time. What do you think is happening? Try to come up with an explanation for the behaviour you see.
Another end point /count returns a count of how many GET requests the server has processed. Try spamming our NodePort with requests to /count like for i in $(seq 1 4000); do curl <IP address here>/count; done. What do you notice about the count for each server? Try to explain what you see.

Hint: Think all the way back to the second question.
Try scaling up the replication count of comp4000server. You can do this using kubectl scale deployment comp4000server --replicas <n> where <n> is a number of your choosing. Try this a few different times with different numbers and examine the output of kubectl get pods each time. You might also want to try repeating some of the earlier exercises with your newly scaled deployment.

Part 2: Deploying a Custom Container Image (Hard)

The simple web server for this experience has one last end point: /printerfacts. Making an HTTP GET request to this end point returns some totally non-suspicious facts about printers. Your task is to write a simple client application (in any language of your choosing) to consume the printerfacts API in some way. Feel free to be as creative as you like.

Once you have written your application, containerize it and deploy it to your cluster in any way you see fit. For example, you may wish to replicate it, make it stateful (e.g. persist a simple database), or expose it to the outside world using a LoadBalancer service.

Tell us about what you did and how you did it. In particular, we want to hear about the parts you had difficulty with, any ideas you had that didn’t pan out, what ended up working, and how your deployment fits together with the rest of the cluster.

Points for this question will be awarded based on:

The sophistication of your deployment. (Challenge yourself!)
The quality of your explanation.

Here are some hints to help you get started:

When you are creating your container image, you need to take extra care to ensure Kubernetes uses your local image. To do this, run eval $(minikube docker-env) before running any docker commands in your shell. In your deployment.yml, set imagePullPolicy: Never under the container field to force it to use your local image.
You can create a new container image by writing a Dockerfile for it and running docker build -t <name> . to build it.
You can modify the deployment.yml file to create your API object, then re-deploy it by running the same kubectl apply -f command from earlier. Feel free to copy-paste and/or modify any of the existing configuration.

Part 3: Reflection

Summarize your experience with Kubernetes in a few paragraphs (both the good and the bad). What concepts do you see reflected here from the research papers we have read thus far? After having some hands on experience with a distributed system technology, have any of your opinions or initial assumptions changed? Feel free to list any other thoughts you have.

Acknowledgements

The idea for the printerfacts API comes from Christine Dodrill’s wonderful blog post.