[BEAM-6018] Memory leak in GCSUtil.java executeBatches

• Type: Bug
• Status: Resolved
• Priority: P2
• Resolution: Fixed
• Affects Version/s: 2.7.0
• Fix Version/s: 2.9.0
• Component/s: io-java-gcp
• Assignee: Udi Meiri
• Reporter: Udi Meiri
• Created: 08/Nov/18 01:59
• Updated: 16/May/20 13:52
• Resolved: 14/Nov/18 21:14

Description

In streaming pipelines there are multiple calls to moveToOutputFiles (https://github.com/apache/beam/blob/42984a821b3e73aee2966d11d7fb436b5ff22b68/sdks/java/core/src/main/java/org/apache/beam/sdk/io/WriteFiles.java#L796).
When writing to GCS, this call uses executeBatches (https://github.com/apache/beam/blob/279a05604b83a54e8e5a79e13d8761f94841f326/sdks/java/extensions/google-cloud-platform-core/src/main/java/org/apache/beam/sdk/util/GcsUtil.java#L551), which wraps a thread pool in MoreExecutors.getExitingExecutorService(). This wrapper introduces a DelayedShutdownHook which persists until the worker exits.

Issue Links

• links to: GitHub Pull Request #6994