apache / apache/arrow-java

[Java] MacOS SIGABRT in ORC JNI adapter when destructing multiple ORC readers

未關閉
#60 5 則留言 0 個 reaction 已指派 0 人 在 GitHub 檢視
Type: bug
主要語言
Java
星號
94
分支
152
平均合併
3 天 16 小時
30 天內合併 PR
11

描述

### Describe the bug, including details regarding any error messages, version, and platform.

I was playing around with the Arrow Java ORC adapter and noticed an intermittent SIGABRT when `.clear()`ing ORC file/stripe readers.

On MacOS M1 (aarch64), I can reproduce locally with this diff: https://github.com/apache/arrow/pull/43699. However, this diff does not fail in CI, which is configured to run only on Linux. The test iteratively creates a new ORC file, reads a stripe, then deletes it. The test is repeated 100 times. The SIGABRT is intermittent and typically crashes on a random test iteration.

My assumption is that there is a memory management issue in the Arrow ORC JNI / Arrow ORC / Apache ORC C++ layers that manifests when destructing the ORC C++ objects.

### Component(s)

Java

貢獻指南

開啟貢獻指南

研究方向

從 apache/arrow#43699 中的 reproducer diff 開始,並在 macOS M1 上執行其重複的 ORC file/stripe reader 測試。追蹤 Java ORC JNI、Arrow ORC 和 Apache ORC C++ 的銷毀路徑;完成的標準是 100 次迭代的測試不再間歇性地引發 SIGABRT。

由索引模型根據 Issue 內容生成。

評估

技術堆疊
cpp, java
領域
data-engineering
Issue 類型
缺陷
難度
4/5
預估耗時
3-5 天
活躍度
停滯
描述清晰度
需要釐清
新手友好度
35/100

把新 issue 寄到你的電子郵件信箱

精選適合新手參與的 GitHub issue 摘要。