Build and Deploy Your First Image to Your First Cluster
DigitalOcean Kubernetes (DOKS) is a managed Kubernetes service. Deploy Kubernetes clusters with a fully managed control plane, high availability, autoscaling, and native integration with DigitalOcean Load Balancers and volumes. You can add node pools using shared and dedicated CPUs, and NVIDIA H100 GPUs in a single GPU or 8 GPU configuration. DOKS clusters are compatible with standard Kubernetes toolchains and the DigitalOcean API and CLI.
In this tutorial, you create an app, build it as a Docker image, and run it on a DigitalOcean Kubernetes cluster. You also learn how to use DigitalOcean Container Registry to store your Docker images for use in your cluster.
Prerequisites
To follow this tutorial, you must:
- Install
doctl, the DigitalOcean command-line tool. - Install
kubectl, the Kubernetes command-line tool. - Sign up for a DigitalOcean account.
STEP 1: Authenticate Your Environment
First, generate a token with read and write access, using any name of your choice. The token is only displayed once, so save it somewhere safe.
Use the following command to authenticate doctl. Paste in your token when prompted.
doctl auth init
STEP 2: Build a Sample Python App
Create a sample app that outputs a “Hello World!” message and its hostname to the screen.
In a new directory, create a app.py file with and write the following content:
from flask import Flask
import os
import socket
app = Flask(__name__)
@app.route("/")
def hello():
html = """Hello {name}!
Hostname: {hostname}"""
return html.format(name=os.getenv("NAME", "world"), hostname=socket.gethostname())
if __name__ == "__main__":
app.run(host='0.0.0.0', port=80)
This code uses the Flask web framework to create a web server that listens on port 80. When you visit the server in a browser, it returns a “Hello World!” message and the hostname of the machine serving the request. The NAME environment variable is used to customize the message. If NAME is not set, it defaults to “world”. This variable is set in the next step.
Create a requirements.txt file with the following content. In the next step, the Docker image configuration uses this file to install Flask as the one Python dependency for your app.
Flask
STEP 3: Build a Docker Image
To build a Docker image, you first need to create a Dockerfile. A Dockerfile is a text document that defines the environment, build, and run commands that your code needs to run. Executing a Dockerfile creates a build artifact called an image.
The ability to compose environments this way with a Dockerfile is a significant benefit of using containers. With an image, code is bundled with the environment it needs to run, which solves the problem of getting code to run correctly in different environments, like on different machines.
In the same directory, create a file called Dockerfile and write the following content:
# Use an official Python runtime as a parent image.
FROM python
# Set the working directory to /app
WORKDIR /app
# Copy the current directory contents into the container at /app
ADD . /app
# Install any needed packages specified in requirements.txt
RUN pip install -r requirements.txt
# Make port 80 available to the world outside this container
EXPOSE 80
# Define environment variable
ENV NAME World
# Run app.py when the container launches
CMD ["python", "app.py"]
The commands in this Dockerfile define an environment with Python set up and Flask installed via pip, move the app’s code to where Flask expects it, define the NAME environment variable that the app needs, open HTTP port 80 to the world, and define the command that runs your app.
Run the following command to build an image based on the Dockerfile you defined. Tag it with -t so it has a human-readable name:
docker build -t my-python-app .
Note
--platform flag in the docker build command or set the default platform in the DOCKER_DEFAULT_PLATFORM environment variable. For more information, see the Docker documentation.Once built, Docker images are stored in a registry, which is a collection of images. You can use the docker images command to see the images you have locally.
docker images
Use the following command to run the image. The -ti arguments ensure you can enter CTRL/CMD+C to exit, and the -p argument maps port 80 of the container to port 80 of the host machine, so you can reach the web server from your browser:
docker run -ti -p 80:80 my-python-app
Go to http://localhost in your browser to see the output of your app. It displays the “Hello World!” message and the hostname of the machine serving the request. You are now accessing your app from a container, which is a runtime instance of the image. Docker returns the ID of the container for a hostname, simulating how a virtual machine might slot in the machine hostname. Therefore, each time you run the image, you get a new container, and therefore a new hostname.
Enter CTRL/CMD+C in your terminal to exit the app. Then, run the app again:
docker run -ti -p 80:80 my-python-app
The hostname changes, demonstrating that the image is the immutable artifact that contains your app, and the container is an ephemeral runtime instance of that image.
STEP 4: Upload Your Docker Image
The image is still only in your local registry. To make it available to other machines, you need to upload it to a registry server.
Since you have already authenticated your environment with your DigitalOcean account, you can create such a registry with DigitalOcean Container Registry (DOCR). Run the following command using your registry name for the <your-registry-name> variable:
doctl registry create <your-registry-name>
Then, log into your registry with this commend. This configures Docker to use DOCR as the default registry for pushing and pulling images:
doctl registry login
Next, tag your image with the fully-qualified name of your registry, which tells Docker where to upload it:
docker tag my-python-app registry.digitalocean.com/<your-registry-name>/my-python-app
Finally, push your image to the registry:
docker push registry.digitalocean.com/<your-registry-name>/my-python-app
Once uploaded, you can use the following command to create a running container of your app on any machine that you have authenticated your environment on (with doctl registry login):
docker run -p 80:80 registry.digitalocean.com/<your-registry-name>/my-python-app
STEP 5: Create a Cluster
Now that you have a containerized version of your app stored in a cloud registry, you can run it on Kubernetes at scale using DigitalOcean Kubernetes, which provides a management layer that offers convenience features like automatic upgrades, monitoring, security patches, metrics, logging, and a user interface for managing your cluster.
From one image you can create many containers, and to run at scale, you need something like Kubernetes to manage them, which is an orchestrator. An orchestrator is a management application that coordinates running containers on a cluster of machines. Each machine in a cluster is called a node. In DigitalOcean Kubernetes, these nodes are Droplets. If your app needs more capacity, you can add more nodes to your cluster. Kubernetes then packs them efficiently with your containers, managing their state and secrets for you.
DigitalOcean Kubernetes gives users the ability to manage groups of nodes in tandem using node pools, which are groups of nodes that you configure to be of a certain size and number. If enabled, node pools auto-scale, increasing or decreasing the number of nodes when their resources are over or underutilized. You can also configure node pools to automatically upgrade their nodes to the latest patch version of Kubernetes.
Create a new DigitalOcean Kubernetes cluster with the following command:
doctl kubernetes cluster create <your-cluster-name> --tag do-tutorial --auto-upgrade=true --node-pool "name=mypool;count=2;auto-scale=true;min-nodes=1;max-nodes=3;tag=do-tutorial"
This command creates a cluster with one node pool named mypool featuring two nodes to start. It also allows the cluster to automatically scale the node pool in size between one and three nodes (depending on the needed capacity). Finally, it tags everything with do-tutorial, and enables automatic upgrades to keep the cluster up to date with security patches and upgrades.
Note
doctl automatically configures the Kubernetes command-line interface, kubectl, so that all kubectl commands execute against your new cluster. This is similar to how doctl registry login configured the docker CLI to use your new registry. Therefore, you can use the kubectl commands in this tutorial to manage your cluster without further configuration. If you need to manage a different cluster, you can change kubectl’s context using doctl kubernetes cluster kubeconfig.There are a number of machine sizes you can use for a node, each one offering a different combination of memory and CPU cores. When you create a node pool, you can configure machine size using any slug you see with doctl kubernetes options sizes. If you change the desired machine size after creating the node pool, DigitalOcean Kubernetes recycles the nodes, which destroys the old nodes at the same rate as they are being replaced with the new ones.
STEP 6: Run Your App on a Cluster
Your cluster is ready when you get output that looks like this:
Notice: Cluster is provisioning, waiting for cluster to be running
......................................................
Notice: Cluster created, fetching credentials
Notice: Adding cluster credentials to kubeconfig file found in "/root/.kube/config"
Notice: Setting current-context to do-nyc1-*********
The components of a Kubernetes cluster are defined by YAML files called manifests. A common Kubernetes workflow is a cycle of defining or modifying a manifest and then applying it to a cluster with kubectl apply.
Here, the next step is authorizing your cluster to use your private registry so it can download images. You can use doctl to generate the manifest for this, and then pipe the generated manifest directly to kubectl to apply it:
doctl registry kubernetes-manifest | kubectl apply -f -
Kubernetes stores your registry credentials as a secret, the built-in mechanism Kubernetes offers for securely storing sensitive data. After running this command, the secret is uploaded and given a name similar to your registry’s name.
Next, configure Kubernetes to use this secret as an authentication token when pulling images from your private registry:
kubectl patch serviceaccount default -p '{"imagePullSecrets": [{"name": "registry-<your-registry-name>"}]}'
Finally, you can deploy the app. To create a Deployment of your app, which is the object Kubernetes uses to maintain the desired state of your running containers, run the following command. This launches the app live in the cluster:
kubectl create deployment my-python-app --image=registry.digitalocean.com/<your-registry-name>/my-python-app
Each deployment creates a replica set, which is the object Kubernetes uses to maintain a stable number of replicas of your container. Each replica is a separate running instance of your container called a pod. Replica sets track the desired number of pods you wish to run for your app. By default, this is set to one, as you see when you run this command:
kubectl get rs
NAME DESIRED CURRENT READY AGE
my-python-app-84b997f5b4 1 1 1 5s
To view details at the pod level, run the following command:
kubectl get pods
NAME READY STATUS RESTARTS AGE
my-python-app-84b997f5b4-6j5pn 1/1 Running 0 27s
Try scaling up to run 20 replicas:
kubectl scale deployment/my-python-app --replicas=20
Now, when you call kubectl get rs and kubectl get pods, you see a lot more activity. If you repeatedly call kubectl get pods, you can watch the Status column change as Kubernetes gets the 19 new pods up and running.
Next, run the following command to see how these Pods get divided over your nodes:
kubectl get pod -o=custom-columns=NODE:.spec.nodeName,NAME:.metadata.name --all-namespaces | grep my-python-app
This returns an output similar to the following:
mypool-byrky my-python-app-84b997f5b4-25shx
mypool-byrky my-python-app-84b997f5b4-2tkgz
mypool-zrkyz my-python-app-84b997f5b4-5dtbz
mypool-zrkyz my-python-app-84b997f5b4-5gl7h
...
Since you named your node pool mypool, the two individual nodes have names with mypool, plus some random characters. Furthermore, the pods are being scheduled so that they are comfortably spread out on your available capacity.
Next, create a load balancer to expose your deployment to the world. The load balancer runs in the cloud and routes the incoming traffic:
kubectl expose deployment my-python-app --type=LoadBalancer --port=80 --target-port=80
This command exposes the pods to the world behind a load balancer, which receives traffic at port 80 and routes that traffic to port 80 on the pods, mapping the host and container ports similarly to how the -p flag was used when testing the app with Docker.
Keep running this command until you see active under the Status column for the new load balancer:
doctl compute load-balancer list --format Name,Created,IP,Status
This returns an output similar to the following:
Name Created At IP Status
a55a6520a74d5437fa389891f2f8708f 2022-04-27T14:46:34Z 143.244.215.xxx new
Status is new and the IP is blank; re-run the command until you have been assigned an IP address.Navigate to the IP address of the load balancer and refresh your browser. The hostname you used earlier changes with every refresh, cycling between the container IDs. This confirms that you have multiple healthy Pods running and serving traffic in a load-balanced way.
STEP 7: Check the Kubernetes Dashboard
You can use the Kubernetes dashboard to inspect your cluster and view visualizers and graphs of your cluster’s state. Install the 1-Click App and then port-forward the dashboard to your local machine to access it.
Summary
You created a sample app, built a Docker image of it, created a private registry, uploaded your image, created a cluster, deployed your application to it, scaled your app up 20x, and exposed it to the world over a load balancer, making it accessible at a stable IP address.
What’s Next
Now that you know the basics, we recommend reading the Kubernetes Operational Readiness Guide to learn how to set up tools such as nginx and Cert-Manager to make your cluster operationally ready to serve production traffic over HTTPS.
We also recommend that you set up push-to-deploy with the official DigitalOcean GitHub Action to automatically build and deploy your code to your cluster when you push to your GitHub repository.
Finally, we recommend learning how to configure auto-scaling for your cluster when resources are over or underutilized.