Skip to content

Add frequent questions to failure notice.

Alex Ives requested to merge alexives/add_faq_to_failure_notice into master

I added some frequent questions to the pipeline failure notice.

Looks something like:

❌ Error: Database testing pipeline failure

Failed pipeline ID: <%= ENV['CI_PIPELINE_ID'] %>

The database testing pipeline has failed for pipeline ID 12345. Please reach out in #database or to a database maintainer to help troubleshoot.

Why aren't details of the failure posted here?

Because migration testing pipelines use production data, and this merge request is public. Production data may contain sensitive information, so we only expose this data in job artifacts and output from the pipeline itself.

Why don't I have access to this pipeline?

Because migration testing pipelines use production data, sensitive information could be leaked in job logs. As a result, the reports are limited to:

Can I just re-run the pipeline?

Possibly yes, if the errors were caused by an incident (or services being slow) and re-running could fix it. If you're doing something slow, remember:

  • Migration testing jobs time out after 10 hours. Synchronous work on a big table can reach this limit, and re-running will mean you wait another 10 hours for it.
  • Migration sampling runs 30 minutes per background migration. If you add 20 background migrations, you will hit the timeout. Consider breaking multiple background migrations apart into separate merge requests.

If you think re-running might fix it, re-trigger the pipeline by running the manual job db:gitlabcom-database-testing.

Edited by Alex Ives

Merge request reports