[
https://issues.apache.org/jira/browse/PARQUET-1470?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16717272#comment-16717272
]
Arnaud Linz commented on PARQUET-1470:
--
I tried, but I'm not a regular commiter and don't have push access to the repo.
It would be quicker if someone else takes care of it.
> Inputstream leakage in ParquetFileWriter.appendFile
> ---
>
> Key: PARQUET-1470
> URL: https://issues.apache.org/jira/browse/PARQUET-1470
> Project: Parquet
> Issue Type: Bug
> Components: parquet-mr
>Affects Versions: 1.10.0
>Reporter: Arnaud Linz
>Priority: Major
>
> Current implementation of ParquetFileWriter.appendFile is:
>
> {{public void appendFile(InputFile file) throws IOException {}}
> {{ ParquetFileReader.open(file).appendTo(this);}}
> {{ }}}
> this method never closes the inputstream created when the file is opened in
> the ParquetFileReader constructor.
> This leads for instance to TooManyFilesOpened exceptions when large merge are
> made with the parquet tools.
> something like
> {{ try (ParquetFileReader reader = ParquetFileReader.open(file)) {}}
> {{ reader.appendTo(this);}}
> {{ }}}
> would be cleaner.
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
