[SPARK-47565] PySpark workers dying in daemon mode idle queue fail query

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 3.4.2, 3.5.1, 3.3.4
• Fix Version/s: 4.0.0
• Component/s: PySpark
• Labels: pull-request-available
• Assignee: Nikita Awasthi
• Reporter: Sebastian Hillig
• Shepherd: Hyukjin Kwon
• Created: 26/Mar/24 12:28
• Updated: 04/Apr/24 11:16
• Resolved: 04/Apr/24 11:16

Description

PySpark workers may die after entering the idle queue in `PythonWorkerFactory`. This may happen because of code that runs in the process, or external factors.

When drawn from the warmpool, such a worker will result in an I/O exception on the first read/write .

Issue Links

• links to: [Github] Pull Request #45635 (sebastianhillig-db)
• links to: GitHub Pull Request #45635