Testdata design, curation and cleanup

We need to start organizing the test data the framework outputs for better readability, ease of debugging and cleanup.

Problem

• Readability The test data created from GitLab QA is very hard to navigate around https://gitlab.com/gitlab-qa-sandbox-group
  • Generic name and timestamps do not help with readability.
• Reusability Tests are not lean in setup, we liberally create groups and projects. A number of these can be reused.
  • See: gitlab-org/gitlab-qa#307 (closed)
• Stability Tests are failing on envs which we don't clean up test data. Tests from canary and production pipelines have been failing because of the number of projects/groups created.
  • We recently cleared out https://gitlab.com/gitlab-qa-sandbox-group and a few days later we now have 41 pages of groups. https://gitlab.com/gitlab-qa-sandbox-group?page=41
  • We need to periodically clean out the test data on staging and also canary.

Proposal

• Define isolation levels of tests data. For tests that are in a lower level it can reuse a static parent container.
  1. Group
     1. Project
        1. Issue
        2. Merge requests.
• Design and separate write and readonly test data. Organize the test data for ease of nagivation.
  • Readonly data: Big MRs/issues that the test just needs to navigate to and view the UI.
  • Separate test data into stage areas (dev,create,plan) or etc
• Add inteligence into the test. Use existing groups and projects if created already
• Name artifacts the test create after the test name or what is being tested. Instead of creating (awesome project) the project should be named after what is being tested with a timestamp.
  • awesome-project-a52842b18041e9b9 => create-project-test-20190201-1122334