[BEAM-8151] Allow the Python SDK to use many many threads

• Type: Improvement
• Status: Resolved
• Priority: P2
• Resolution: Fixed
• Fix Version/s: 2.18.0
• Component/s: sdk-py-core, sdk-py-harness
• Assignee: Luke Cwik
• Reporter: Luke Cwik
• Created: 04/Sep/19 19:07
• Updated: 16/May/20 14:13
• Resolved: 15/Nov/19 17:56

Description

We need to use a thread pool which shrinks the number of active threads when they are not being used.

 

This is to prevent any stuckness issues related to a runner scheduling more work items then there are "work" threads inside the SDK harness.

 

By default the control plane should have all "requests" being processed in parallel and the runner is responsible for not overloading the SDK with too much work.

Issue Links

• causes: BEAM-8944 Python SDK harness performance degradation with UnboundedThreadPoolExecutor
• is duplicated by: BEAM-8617 Tear down the DoFns upon the control service termination in Python SDK harness
• links to: GitHub Pull Request #9477
• links to: GitHub Pull Request #10091
• links to: GitHub Pull Request #10123
• links to: GitHub Pull Request #10134