[BEAM-5719] Misconfigured GCS staging jobs can nuke entire top-level bucket

• Type: Sub-task
• Status: Resolved
• Priority: P2
• Resolution: Fixed
• Fix Version/s: Not applicable
• Component/s: build-system, website
• Assignee: Scott Wegner
• Reporter: Scott Wegner
• Created: 11/Oct/18 18:44
• Updated: 16/May/20 13:33
• Resolved: 11/Oct/18 22:22

Description

See: https://github.com/apache/beam/pull/6655#issuecomment-429037632

A bug in the Website_Stage_GCS job was causing the Jenkins job to wipe out the top-level GCS bucket used to stage and serve the website from. This ends up blocking all future runs of the job and takes manual effort to recreate and configure the bucket.

The linked PR should fix the immediate issue, but it's still the case that a bug in the job/configuration could potentially wipe it out again. We should refactor the job or modify the command so that there is no longer danger of wiping it out.

Issue Links

• links to: GitHub Pull Request #6660