You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@zeppelin.apache.org by "ChoiByungHo (Jira)" <ji...@apache.org> on 2025/07/24 15:03:00 UTC
[jira] [Created] (ZEPPELIN-6258) Improve Process resource management in SparkInterpreterLauncher
ChoiByungHo created ZEPPELIN-6258:
-------------------------------------
Summary: Improve Process resource management in SparkInterpreterLauncher
Key: ZEPPELIN-6258
URL: https://issues.apache.org/jira/browse/ZEPPELIN-6258
Project: Zeppelin
Issue Type: Bug
Reporter: ChoiByungHo
Assignee: ChoiByungHo
The `detectSparkScalaVersion` method in `SparkInterpreterLauncher.java` has potential issues with Process resource management that could lead to process hangs and resource leaks.
Current Issues
1. Missing stdout stream handling
The process's stdout stream is not consumed, which can cause buffer overflow and process hang if the process produces output to stdout.
2. No timeout for process execution
The `process.waitFor()` call has no timeout, potentially causing indefinite blocking if spark-submit hangs.
3. Process not properly destroyed
The Process instance is not explicitly destroyed after use, which may leave zombie processes.
4. No exit value validation
The process exit value is not checked, making it difficult to diagnose failures.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)