Octopus Deploy Documentation

Execution containers for workers

Last updated

For a step running on a worker or on the Octopus Server, you can select a Docker image to execute the step inside of.

When an execution container is configured for a step, Octopus will still connect to the worker machine via a Tentacle or SSH. The difference is that the specified image will be run as a container and the step will be executed inside the container.

See the blog post announcing this feature for some added context.

Requirements

You need Docker installed and running on the worker/Octopus Server (built-in worker), in order to use execution containers for workers

Octopus cloud dynamic worker pools

Octopus Cloud dynamic workers have Docker pre-installed and support execution containers, with the exception of Windows 2016 pools. Dynamic worker pools with a VM type of Windows Server Core 2016 do not have Docker installed, and they cannot be used to run execution containers.

How to use execution containers for workers

  • Configure a feed in Octopus Deploy for a Docker registry.
  • Add a project and define a deployment process (or add a runbook).
  • Set the Execution Location for your step to Run on a worker.
  • In Container Image select Runs on a worker inside a container.
  • Choose the previously added container registry.
  • Enter the name of the image (aka execution container) you want your step to run in. (e.g. octopusdeploy/worker-tools:3.2.0-windows.ltsc2019).
  • Click Save.
  • Click Create release & deploy.

First deployment on a docker container

Pre-pulling your chosen image will save you time during deployments.

When you choose to run one or more of your deployment steps in a container, your deployment process will docker pull the image you provide at the start of each deployment during package acquisition.

For your first deployment this may take a while since your docker image won't be cached. You can pre-pull the desired docker image on your worker before your first deployment to avoid any delays.

Which Docker images can I use?

The easiest way to get started is to use the worker-tools images built by Octopus Deploy.

When a step is configured to use an execution container, you can choose from:

The octopusdeploy/worker-tools Docker images

For convenience, we provide some images on Docker Hub octopusdeploy/worker-tools which include common tools used in deployments.

We recommend using our worker-tools image as a starting point for your own custom image to run on a worker.

The canonical source for what is contained in the octopusdeploy/worker-tools images is the Dockerfile's in the GitHub repo. For example:

Some of the tools included are:

  • Octopus Deploy CLI and .NET client library
  • .NET Core
  • Java (JDK)
  • NodeJS
  • Azure CLI
  • AWS CLI
  • Google Cloud CLI
  • kubectl
  • Helm
  • Terraform
  • Python

Custom docker images

It can be beneficial to build your own custom Docker image when using execution containers, particularly when you wish the image size to be as small as possible.

Supported Windows base images

For Windows images, we recommend using a base image no older than the ltsc2019 image. Octopus does not support images that are older than ltsc2019, and while containers based on these images can still run steps, you may run into unexpected issues.

If your containers are based on an earlier image of Windows, we strongly recommend upgrading your workers to Windows 2019 and rebasing your Docker containers to use a 2019 base image.

Supported Linux distributions

It's important to understand there are some limits to which Linux Docker images can be used as a container image. The Docker image must be based on a Linux distribution using the GNU C library, or glibc. This includes operating systems like Ubuntu, Debian, and Fedora.

Linux distributions built on muscl, most notably Alpine, do not support Calamari, and cannot be used as a container image. This is due to Calamari currently only being compiled against glibc and not musl.

You can usually find the base operating system of a Linux Docker image by running the following command:

docker run --entrypoint='' [image name] /bin/cat /etc/os-release.

For example for the octopusdeploy/worker-tools:2.0.1-ubuntu.18.04 image, you'd run:

docker run --entrypoint='' octopusdeploy/worker-tools:2.0.1-ubuntu.18.04 /bin/cat /etc/os-release.

Required OS dependencies

The operating system must also include a number of dependencies required to support .NET Core applications. When a step is configured to use an execution container, Calamari (the Octopus deployment utility) is executed inside the specified container. Since Calamari is a .NET Core self-contained executable, any custom Docker image needs to include the dependencies required to execute a .NET self-contained executable.

The Microsoft .NET Core documentation lists the dependencies required for a .NET Core application with popular Linux distributions.

If a third party container is missing a library, it is usually the libicu library. The error Couldn't find a valid ICU package installed on the system indicates the ICU library is missing.

If your chosen Docker image does not have these prerequisites, the easiest solution is to create a custom Docker image based on the image you wish to use, install the required libraries, push the image to a repository like DockerHub, and select your custom image as the container image.

Microsoft also provides base images that include these dependencies.

Custom docker image example

The following example is a basic Docker file that (when built) can run Calamari and PowerShell scripts:

FROM ubuntu:20.04

ARG POWERSHELL_VERSION=7.1.3\*

RUN apt-get update && \
    apt-get install -y --no-install-recommends \
        curl \
        unzip \
        apt-transport-https \
        software-properties-common && \
    apt-get clean && \
    rm -rf /var/lib/apt/lists/*

# Powershell core
# https://docs.microsoft.com/en-us/powershell/scripting/install/installing-powershell-core-on-linux?view=powershell-7.1#ubuntu-2004
RUN curl -LO -k "https://packages.microsoft.com/config/ubuntu/20.04/packages-microsoft-prod.deb" && \
    dpkg -i packages-microsoft-prod.deb && \
    apt-get update && \
    add-apt-repository universe && \
    apt-get install -y --no-install-recommends \
        powershell=${POWERSHELL_VERSION} && \
    apt-get clean && \
    rm -rf /var/lib/apt/lists/* && \
    rm -f packages-microsoft-prod.deb

To learn more about creating a custom docker image, we have a detailed blog post that describes how to get started and the minimum set of dependencies you would need.

Tool paths

Because Calamari is executed directly inside the specified container, execution containers on workers are run in non-interactive mode. Since the execution container is not running interactively, it does not process your .bashrc file. If the tool you have installed relies on .bashrc to modify the path (e.g. nvm) to include a non-standard folder, you will need to manually define the additional directories in the $PATH variable in your Dockerfile using the ENV PATH directive.

For example, if you install node.js via nvm, you will need to remediate the $PATH variable in your image with the location node.js gets installed to with the following directive in your Dockerfile:

ENV PATH="/root/.nvm/versions/node/v${NODE_VERSION}/bin:$PATH"`

CMD and ENTRYPOINT directives

Docker images used with the execution containers feature require that no CMD or ENTRYPOINT directives be defined in your Dockerfile.

Including one of these directives will result in the step failing.

Need support? We're here to help.