apache / apache/parquet-java

ParquetWriter::close sometimes fail

Open
#1,701 1 comment 0 reactions 0 assignees View on GitHub
Component: Java Component: Parquet Priority: Major Type: bug
Dominant language
Java
Stars
3.1k
Forks
1.6k
Avg merge
3d 12h
Merged PRs (30d)
33

Description

we sometimes run into an exception when closing a ParquetWriter instance:

 
```java

2024-06-10 10:44:01.398    org.apache.parquet.util.AutoCloseables$ParquetCloseResourceException: Unable to close resource
2024-06-10 10:44:01.398        at org.apache.parquet.util.AutoCloseables.uncheckedClose(AutoCloseables.java:85)
2024-06-10 10:44:01.398        at org.apache.parquet.util.AutoCloseables.uncheckedClose(AutoCloseables.java:94)
2024-06-10 10:44:01.398        at org.apache.parquet.hadoop.InternalParquetRecordWriter.close(InternalParquetRecordWriter.java:144)
2024-06-10 10:44:01.398        at org.apache.parquet.hadoop.ParquetWriter.close(ParquetWriter.java:437)
2024-06-10 10:44:01.398    Caused by: java.nio.channels.ClosedChannelException: null
2024-06-10 10:44:01.398        at org.apache.hadoop.hdfs.ExceptionLastSeen.throwException4Close(ExceptionLastSeen.java:73)
2024-06-10 10:44:01.398        at org.apache.hadoop.hdfs.DFSOutputStream.checkClosed(DFSOutputStream.java:158)
2024-06-10 10:44:01.398        at org.apache.hadoop.hdfs.DFSOutputStream.flushOrSync(DFSOutputStream.java:639)
2024-06-10 10:44:01.398        at org.apache.hadoop.hdfs.DFSOutputStream.hflush(DFSOutputStream.java:594)
2024-06-10 10:44:01.398        at org.apache.hadoop.fs.FSDataOutputStream.hflush(FSDataOutputStream.java:136)
2024-06-10 10:44:01.398        at org.apache.parquet.hadoop.util.HadoopPositionOutputStream.close(HadoopPositionOutputStream.java:65)
2024-06-10 10:44:01.398        at org.apache.parquet.hadoop.ParquetFileWriter.close(ParquetFileWriter.java:1663)
2024-06-10 10:44:01.398        at org.apache.parquet.util.AutoCloseables.close(AutoCloseables.java:49)
2024-06-10 10:44:01.398        at org.apache.parquet.util.AutoCloseables.uncheckedClose(AutoCloseables.java:83)
```

**Reporter**: [Roelof Naude](https://issues.apache.org/jira/secure/ViewProfile.jspa?name=naude-r)

**Note**: *This issue was originally created as [PARQUET-2496](https://issues.apache.org/jira/browse/PARQUET-2496). Please see the [migration documentation](https://issues.apache.org/jira/browse/PARQUET-2502) for further details.*

Contributor guide

No contributing guide indexed for this repository

Research direction

Trace the close path from ParquetWriter.close through InternalParquetRecordWriter.close, ParquetFileWriter.close, and HadoopPositionOutputStream.close, using the supplied ClosedChannelException stack trace as the starting point. Determine the conditions that leave the HDFS output channel closed before hflush, then add coverage for the intermittent failure and verify that closing a ParquetWriter completes without this exception.

Written by the indexing model from the issue text.

Assessment

Tech stack
hadoop, java
Domain
data-engineering
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
32/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.