maedhroz commented on code in PR #4739:
URL: https://github.com/apache/cassandra/pull/4739#discussion_r3074855157
##########
src/java/org/apache/cassandra/db/ReadResponse.java:
##########
@@ -239,10 +247,21 @@ private LocalDataResponse(UnfilteredPartitionIterator
iter, ColumnFilter selecti
private static ByteBuffer build(UnfilteredPartitionIterator iter,
ColumnFilter selection)
{
- try (DataOutputBuffer buffer = new DataOutputBuffer())
+ int initialBufferSize = bufferInitialSizeMin;
+
+ if (bufferInitialSizeMax > bufferInitialSizeMin)
+ {
+ // Size output buffer to 10% above the moving average to
absorb minor variance and limit rebuffering.
+ double estimatedResponseSize = estimatedResponseBytes.get();
+ double bufferSizeEstimate =
Double.isNaN(estimatedResponseSize) ? bufferInitialSizeMin :
estimatedResponseSize * 1.1;
+ initialBufferSize = Math.min((int) bufferSizeEstimate,
bufferInitialSizeMax);
+ }
+
+ try (DataOutputBuffer buffer = new
DataOutputBuffer(initialBufferSize))
{
UnfilteredPartitionIterators.serializerForIntraNode().serialize(iter,
selection, buffer, MessagingService.current_version);
- return buffer.buffer();
+ estimatedResponseBytes.update(buffer.position());
Review Comment:
Given it's not something we can enable at runtime, you're right. I'll
probably leave this though, given we'll enable a larger max by default.
--
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]
