There are no docs about Code Quality disk usage

Summary

Code quality writes does not stop to write data until disk is full.

Steps to reproduce

I followed this documentation : https://docs.gitlab.com/ee/ci/examples/code_quality.html using instruction for latest gitlab-ee stable release 11.3.4-ee.

  • Add a code_quality job as explained in the documentation
  • Run the job

What is the current bug behavior?

The job run until there is no space left on the disk. When I started to run the job, 10 go were available. Then the job processed during 20 minutes and didn't stop to write datas on the disk. In addition, there were not a lot of code to analyze.

As noted by users the Docker in Docker container used by Code Quality can take up quite a bit of disk space and the resulting report can fill a disk as well. The minimum runner requirements to use this functionality aren't documented in the feature documentation.

What is the expected correct behavior?

  • The job should write the report without take so much space.
  • If the job runs out of memory a more graceful exit and/or helpful error message would be nice so users can troubleshoot.
  • Sasha and Devon should know the runner requirements needed to utilize the code quality feature so they can right size their setup.

Relevant logs and/or screenshots

image

Proposal

The documentation for the Code Quality feature should call out the size of the DinD container and/or provide some guidance and/or examples of the disk size needed to utilize the feature. Reference the work done in (#39998 (closed)) for examples.

Edited by James Heimbuck