Composer failing in pipeline but not when manually run

[Mike Holzer]

I am running a Drupal deployment and as part of the build, I have composer pull in dependencies. Since I upgraded to the latest version of Drupal in September, our pipeline has been failing due to a dependency issue. When we run our pipeline manually or (or even just running composer locally), we don't have the dependency issue. 

Is there any way to debug what is going on in the pipeline vs our devs' local environments?

[Theodora Boudale]

Hi Mike,

Do your Pipelines builds run on Atlassian's infrastructure or are you using one of our self-hosted runners?

If the pipelines builds run on Atlassian's infrastructure or with a self-hosted Linux Docker runner, they use Docker containers. For every step of your build, a Docker container starts (the build container) using the image you have specified in your bitbucket-pipelines.yml file. The repo is cloned in the build container, and then the commands of the step's script are executed. When the step finishes that Docker container gets destroyed.

Builds running with self-hosted Linux Shell, MacOS or Windows Runner run directly on the host machine using Bash (for Linux Shell and MacOS runners) and PowerShell (for Windows runners).

It is not possible to debug the build in Pipelines while it is running. You could debug by adding some additional commands in your yml file to run with the next build.

We have the following guide for debugging a build locally with Docker. This is useful if you want to debug a build running on Atlassian's infrastructure or with a self-hosted Linux Docker runner.

• https://confluence.atlassian.com/bbkb/troubleshoot-failed-bitbucket-pipelines-locally-with-docker-1318882094.html

You can try this to see if your build also fails when you debug locally with Docker or if it is successful.

You mentioned that when you run your pipeline manually you don't have this issue, so we would need to see what is different between these two builds.

Does the automated pipeline that fails run on the default, branches, pull-requests, or tags definition?

When you run it manually, do you trigger the pipeline for the same branch/commit as the automated build? And are you selecting the same pipeline definition that failed in the automated run?

Kind regards,
Theodora