Using SpecSync inside a Docker container

If your development pipeline uses Docker containers it might be also useful to run the SpecSync command line tool from a Docker container.

When running SpecSync from Docker containers you can choose from multiple options. The table below summarizes these options. The details of the different options can be found below as sub-sections of this guide. Please also check the general recommendations below.

General recommendations

Mounting vs copying feature files to the container

SpecSync requires to access the feature files and other project and configuration files of your project. In some cases (when linking new scenarios or pulling changes) SpecSync also needs to modify the feature files and these modifications have to be preserved (typically commit and push to Git).

Local modifications can be disabled with the --disableLocalChanges option, that is useful when the synchronization is done in CI/CD pipeline where the changes cannot be preserved.

The most efficient way to access the local files is to define a mounting point (volume) and mount the local project folder when you create the container from the image. This way the changes will be preserved and also the image size can be significantly reduced.

# define a parametrizable mounting point
ARG LOCAL_DIR=/local
# set it as working folder
WORKDIR ${LOCAL_DIR}

The local folders can be mounted to the container when the container is created using the -v <LOCAL-FOLDER>:<MOUNT-POINT> argument of the docker run command. In the images created for SpecSync, the mounting point is usually the /local folder. For example:

docker run -it --rm -v C:\MyProject\src\features:/local myspecsyncimage push

The alternative would be to use the ADD or COPY command, but that essentially creates a copy of your project codebase at the time of building the image. The files copied into the container reduces the reusability of the created image.

With ADD/COPY, the changes that SpecSync makes on the feature files would be lost, therefore the --disableLocalChanges option has to be used.

In the majority of the cases, mounting the local files is a better option, although files that generally remain unchanged (SpecSync license, configuration file) might be added to the image using ADD/COPY.

Set the working directory to the folder of the feature file set instead of the SpecSync tool folder

The default options of SpecSync assume that the current folder is the root of the feature file set (SpecFlow project folder or features folder of a Cucumber project). Typically the SpecSync configuration file (specsync.json) is located in that folder too.

Although it is possible to let SpecSync synchronize feature files from other folders as well or find the configuration elsewhere, it is easier if the current working directory is the folder of the feature file set.

If the project folders are mounted to the image to a folder /local, the working folder should be the /local folder itself or one of its sub-folders.

The following example sets the working folder to the project folder mounting point and sets the default entry point to the SpecSync executable from a different directory (usually /specsync).

WORKDIR /local
ENTRYPOINT [ "/specsync/SpecSync4AzureDevOps" ]

Create a derived Docker image from the official SpecSync Docker image

Creating your own Docker image from the official SpecSync Docker image might be useful when accessing your Azure DevOps installation requires special network configurations (e.g. installing a certificate) or when you would like to simplify the execution parameters (e.g. license file location).

The official SpecSync Docker images are based on the ubuntu:22.04 image and contain the native Linux binaries of SpecSync. Besides that the changes are kept at minimum:

  • the libssl3 and ca-certificates packages are installed using apt-get as these are needed to access the Azure DevOps cloud service.

The official SpecSync Docker image contains the SpecSync binaries in the /specsync folder and expects the local repository to be mounted into the /local folder (see Mounting vs copying feature files to the container). The working folder is set to the /local and the default entry point to /specsync/SpecSync4AzureDevOps.

You can create derived images from the official SpecSync Docker image and customize it. For example you can

  • Install further certificates to access your Azure DevOps installation

  • Include base configuration files (see Hierarchical configuration files)

  • Include license file

  • Include scripts that specify default parameters for SpecSync (e.g. the license file location or user PAT)

You can find further information on how to use such an image in the Install as Docker image page.

Example

The following Dockerfile customizes the official SpecSync Docker image by adding the license file to the image and a shell script that specifies the license file parameter for the SpecSync commands. It also enables logging.

Dockerfile
FROM specsolutions/specsync:latest

# Copy SpecSync license file and the utility script to a /shared folder
WORKDIR /shared
COPY specsync.lic .
COPY runspecsync.sh .
RUN chmod +x ./runspecsync.sh

WORKDIR /local

ENTRYPOINT [ "/shared/runspecsync.sh" ]
runspecsync.sh
#!/bin/bash
/specsync/SpecSync4AzureDevOps $@ --license /shared/specsync.lic --log log.txt

To run SpecSync from the created image, you need to mount the local repository folder and pass further arguments for SpecSync, like

docker run -it --rm -v C:\MyProject\src\features:/local myimage push --tagFilter @foo

Integrate SpecSync into a .NET Core SDK-based image

If you use the Docker container for building or testing .NET Core applications (i.e. you have a .NET Core SDK-based image already), you can easily integrate SpecSync into that as a .NET Core Global tool.

In this setup the easiest is to install SpecSync into the image as a global .NET Core tool. In the Install as .NET Core tool page we recommend installing SpecSync as a local .NET Core Tool to avoid version conflicts between projects on the same machines, but since the Docker container will be bound to one specific project anyway, you can safely use global tools. To install SpecSync as a global .NET Core Tool in the image, you need to add the following line to the Dockerfile.

RUN dotnet tool install --global SpecSync.AzureDevOps --version 3.0.2

Specifying the version is optional. You can even make your image configurable for different SpecSync versions by specifying the SpecSync version as an argument (see example below).

Global .NET Core tools are installed to /root/.dotnet/tools, but this folder is not included in the PATH by default. You can add this folder to the PATH or refer to the SpecSync command line tool as /root/.dotnet/tools/specsync.

Just like with the other Docker container options it is recommended to mount the local folder instead of copying it.

You can build the image to be used for interactive development but you can also populate it with scripts to perform complex tasks, e.g. build project and synchronize test cases with SpecSync push command, or run tests and publish test results with SpecSync.

If you would like to use the contain only to run SpecSync, it is better to use the official containers or build one on pure Linux.

Example

The following Dockerfile creates an image based on the .NET 6 SDK to be able to develop the project interactively.

Dockerfile
FROM mcr.microsoft.com/dotnet/sdk:6.0

ARG SPECSYNC_VERSION=3.3.8
ARG LOCAL_DIR=/local

RUN dotnet tool install --global SpecSync.AzureDevOps --version ${SPECSYNC_VERSION}
ENV PATH="$PATH:/root/.dotnet/tools"

WORKDIR ${LOCAL_DIR}

You can build an image and work with that interactively, e.g.:

> docker run -it --rm -v C:\MyProject\src\features:/local myimage bash
# dotnet build
# specync push --tagFilter @foo

Integrate SpecSync into a pure Linux image

If your project is not .NET based, the most efficient way to include SpecSync into your Docker image is to download the SpecSync native Linux binaries to the image. These binaries contain all pre-requisites, so the .NET framework does not need to be installed to the Docker image.

The SpecSync Linux binaries work with most Linux distributions, but we test it with ubuntu:latest (currently ubuntu:22.04). The Dockerfile should follow the classic download package and unzip it pattern (see example below).

For SpecSync for Azure DevOps before v3.4 the binaries required localization packages by default. This can be fulfilled by installing the necessary ICU packages to the Docker container (e.g. libicu66 on ubuntu:20.04). For newer versions this is not necessary.

For the older versions an alternative to the ICU package installation, the internalization support can be disabled by setting the DOTNET_SYSTEM_GLOBALIZATION_INVARIANT environment variable to 1. (See .NET documentation for details.)

Just like with the other Docker container options it is recommended to mount the local folder instead of copying it.

You can make the image for generic-purpose use (e.g. for interactive work), or you can make it specifically for the purpose of executing the SpecSync command line tool using the ENTRYPOINT command. In case of the latter, you can consider using the official SpecSync Docker images or create a derived image from them.

Example

The following Dockerfile creates an image from ubuntu:22.04, installs the SpecSync native binaries to it and exposes the SpecSync command line tool as a default entry point.

Dockerfile
FROM ubuntu:22.04

RUN apt-get update && apt-get install -y \
    wget \
    unzip \
    && rm -rf /var/lib/apt/lists/*

ARG SPECSYNC_VERSION=3.3.8
ARG LOCAL_DIR=/local

WORKDIR /specsync

RUN wget -qO specsync.zip https://www.specsolutions.eu/media/specsync/SpecSync.AzureDevOps.${SPECSYNC_VERSION}-linux-x64.zip \
  && ( unzip -q specsync.zip || [ $? -le 1 ] ) \
  && rm specsync.zip \
  && bash -c "chmod +x ./SpecSync4AzureDevOps"

# Enable detection of running in a container
ENV DOTNET_RUNNING_IN_CONTAINER=true

WORKDIR ${LOCAL_DIR}

ENTRYPOINT [ "/specsync/SpecSync4AzureDevOps" ]

To run SpecSync from the created image, you need to mount the local repository folder and pass further arguments for SpecSync, like

docker run -it --rm -v C:\MyProject\src\features:/local myimage push --tagFilter @foo

The pure ubuntu image does not contain the packages necessary for HTTPS requests, therefore if your Azure DevOps server needs HTTPS connection, you also need to install the necessary packages. In the example above, we installed the wget package that implicitly included these. If you prepare a package that does not need wget, you need to add the libssl3 and ca-certificates with apt-get.

Last updated