Create and Push a Docker Image From an Octopus Deployment
Creating an Octopus step template to create and push a Docker image to DockerHub is a good way to move from a VM to Containers while keeping the current infrastructure.
Join the DZone community and get the full member experience.Join For Free
Docker is one of the disruptive technologies within virtualization, allowing the different containers to run on the same machine, sharing resources and reducing the overhead. The technology allows DevOps teams to have another tool to develop, build, and ship software. One use case for containers is the Microservices Architecture Pattern.
DockerHub is the image repository offered by Docker where we can publish public repositories (the tech community is there) or use the private repositories feature (instead of building an infrastructure in the local environment). One of the features is the Automated Builds from a GitHub or Bitbucket account. Although it is a great feature, not all software shops uses this set of technologies.
For a software shop using Octopus, there are already features to deploy Octopus images from an Octopus Deployment. However, is not possible to create and push a Docker Image based on packages stored in Octopus. To fill this gap, I created an Octopus step template where it is possible to create and push a Docker image to DockerHub.
This fits in an organization moving from a VM environment to a Container environment, but keeping the current infrastructure (i.e., using Octopus to manage the releases, including where the packages are deployed).
The tools used for this tutorial were:
Visual Studio and the Docker WebApp for the Tutorial
Using the Visual Studio .NET Core Web Application template for this tutorial, the project was created and committed here. The WebApp is the out-of-the-box web application.
Continuous Integration Server
You can use your preferred CI server. In your CI server will need to:
- Publish the Docker WebApp.NET project:
dotnet publish WebAppDemo -c Release
- Create the NuGet package for Octopus:
Octo.exe pack --id DockerWebAppDemo --version 1.0.0
- Upload the AWS Lambda NuGet package to Octopus Server:
NuGet.exe push DockerWebAppDemo.1.0.0.nupkg -ApiKey myApiKey -Source https://myOctopusServer
These instructions are generic, and depending on your CI server technology, can be done in different ways. You can take them as generic steps and adapt to your needs.
Create and Push a Docker Image Octopus Step Template
The step template is simple and straightforward. The Powershell script:
- Get all the required parameters, such as the DockerHub username, password, and image name, among others.
- Validate the parameters, i.e., if they have a value.
- Creates the Docker image.
- Pushes the Docker image.
- Feedback the user.
The step template has a prerequisite; it depends on the Docker for Windows. You need to download and install it on the machines where the step will run.
To create the Octopus image, we need to use a pivot machine with Docker for Windows installed.
We will use the Octopus out-of-the-box features to deploy the DockerWebAppDemo package to the Pivot Machine, and the new step template to create and push the Docker image to DockerHub.
The Octopus Project will look like this:
The first step deploys the NuGet DockerWebAppDemo package to a custom location in the Pivot Machine (with the Docker Machine role).
The second step creates and pushes the Docker image to DockerHub using the Dockerfile provided.
After creating a release (potentially triggered by the CI server), Octopus Server can deploy it.
If the step runs successfully, we can check the Docker image in the repository.
We can even run it in a machine running Docker using the following command:
docker run -p 5000:80 -e "ASPNETCORE_URLS=http://+:80" -it --rm joaoasrosa/dockerwebappdemo
If we access to a browser using the URL:port for the Docker container, we can see the application running.
The step allows the creation of Docker images from an Octopus deployment, centralizing and minimizing the packages maintenance operations. Also, it is possible to deploy the same package to a machine or a Docker image, increasing the deployment scenarios for a DevOps team.
Published at DZone with permission of João Rosa, DZone MVB. See the original article here.
Opinions expressed by DZone contributors are their own.