[FLINK-10383] Hadoop configurations on the classpath seep into the S3 file system configs

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.6.1
• Fix Version/s: 1.6.2, 1.7.0
• Component/s: FileSystems
• Labels: pull-request-available
• Assignee: Stephan Ewen
• Reporter: Stephan Ewen
• Created: 20/Sep/18 17:43
• Updated: 25/Sep/18 07:27
• Resolved: 25/Sep/18 07:25

Description

The S3 connectors are based on a self-contained shaded Hadoop. By design, they should only use config value from the Flink configuration.

However, because Hadoop loads implicitly configs from the classpath, existing "core-site.xml" files can interfere with the configuration in ways intransparent for the user. We should ensure such configs are not loaded.

Issue Links

• links to: GitHub Pull Request #6728