GitHub - Track imports via UI versus API

To improve imports observability, we could add an import_origin column to the projects table, and start recording whether an import is initiated via ui or api. This in combination with the existing import_type column could be useful for metrics.

Options

  1. Adding columns to the db table OR
  2. Using internal events https://docs.gitlab.com/development/internal_analytics/internal_event_instrumentation/

NOTE: Unfortunately we won't be able to add to the projects table, so option 2 might be better.

Tasks (if implementing Option 1)

Database migration:

  • Add import_origin column to projects table
  • Type: string (nullable, no default)
  • Expected values: 'ui', 'api', NULL (for existing/unknown)

app/services/import/github_service.rb

  • Modify execute method to accept origin parameter
  • Pass origin through to project creation

lib/gitlab/legacy_github_import/project_creator.rb

  • Accept origin parameter and set it on the project

app/controllers/import/github_controller.rb

  • Update create action to pass origin: 'ui' to Import::GithubService
  • (Gitea inherits this automatically? Verify this)

lib/api/import_github.rb

  • Update post 'import/github' to pass origin: 'api' to Import::GithubService

specs

  • Update Import::GithubService specs to verify origin is set
  • Update controller specs to verify origin: 'ui' is passed
  • Update API specs to verify origin: 'api' is passed

Follow ups

  • Direct transfer / bulk imports
  • other importers
Edited by Oiza Baiye