Skip to content

Parquet: avoid premature closing of FSDataInputStream object #3276

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
velctor wants to merge 2 commits into from
Closed

Parquet: avoid premature closing of FSDataInputStream object #3276

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
eed9bc2
Parquet: avoid premature closing of FSDataInputStream object
Oct 11, 2021
f9a425a
Merge branch 'apache:master' into avoid-stream-closed
velctor Oct 12, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 4 additions & 1 deletion parquet/src/main/java/org/apache/iceberg/parquet/ParquetIO.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -166,9 +166,11 @@ public long defaultBlockSize() {

private static class ParquetInputFile implements InputFile {
private final org.apache.iceberg.io.InputFile file;
private org.apache.iceberg.io.SeekableInputStream icebergHadoopStream;

private ParquetInputFile(org.apache.iceberg.io.InputFile file) {
this.file = file;
this.icebergHadoopStream = null;
}

@Override
Expand All @@ -178,7 +180,8 @@ public long getLength() throws IOException {

@Override
public SeekableInputStream newStream() throws IOException {
return stream(file.newStream());
icebergHadoopStream = file.newStream();
return stream(icebergHadoopStream);
Copy link
Copy Markdown
Contributor

@rdblue rdblue Oct 19, 2021

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This doesn't look correct to me. The InputFile is a factory interface and is not responsible for keeping references to the streams that are created.

Copy link
Copy Markdown
Author

@velctor velctor Oct 20, 2021 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I know that, but the org.apache.iceberg.io.SeekableStream object will be GC after this.f = file.newStream(); in org.apache.parquet.hadoop.ParquetFileReader class. It is difficult to modify the code in the parquet library, so this may be an easier way to solve this problem, under the premise of keeping #90. I want to know if you have any other ideas.

}
}
}