jojochuang commented on code in PR #5980:
URL: https://github.com/apache/ozone/pull/5980#discussion_r1533045332
##########
hadoop-hdds/client/src/main/java/org/apache/hadoop/hdds/scm/storage/BlockOutputStream.java:
##########
@@ -594,14 +608,24 @@ private void handleFlushInternal(boolean close)
if (totalDataFlushedLength < writtenDataLength) {
refreshCurrentBuffer();
Preconditions.checkArgument(currentBuffer.position() > 0);
- if (currentBuffer.hasRemaining()) {
- writeChunk(currentBuffer);
- }
+
// This can be a partially filled chunk. Since we are flushing the buffer
// here, we just limit this buffer to the current position. So that next
// write will happen in new buffer
- updateFlushLength();
- executePutBlock(close, false);
+ if (currentBuffer.hasRemaining()) {
+ if (writtenDataLength - totalDataFlushedLength < 100 * 1024 &&
+ allowPutBlockPiggybacking) {
+ updateFlushLength();
+ writeSmallChunk(currentBuffer);
+ } else {
+ writeChunk(currentBuffer);
+ updateFlushLength();
+ executePutBlock(close, false);
+ }
+ } else {
+ updateFlushLength();
+ executePutBlock(close, false);
Review Comment:
This is the typical case (no hflush) where a PutBlock is sent to update
metadata after four 1-MB chunks are sent via WriteChunk requests)
--
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]
