wgtmac commented on code in PR #3559:
URL: https://github.com/apache/parquet-java/pull/3559#discussion_r3643419818
##########
parquet-hadoop/src/main/java/org/apache/parquet/hadoop/ParquetFileReader.java:
##########
@@ -2368,6 +2384,10 @@ public void readFromVectoredRange(ParquetFileRange
currRange, ChunkListBuilder b
LOG.error(error, e);
throw new IOException(error, e);
}
+ // Release the vectored-read buffer back to the allocator when the row
group is closed.
+ // Requires fs.file.checksum.verify=false so the returned buffer is the
allocator buffer
+ // rather than a sliced subset (see Hadoop's fs.file.checksum.verify
docs).
+ builder.addBuffersToRelease(Collections.singletonList(buffer));
Review Comment:
Please register the buffer immediately after `awaitFuture` succeeds.
`setReadMetrics` invokes user callback code before the buffer is registered, so
an exception there leaks the current buffer despite the outer cleanup.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
