danny0405 commented on a change in pull request #2923:
URL: https://github.com/apache/hudi/pull/2923#discussion_r647059387
##########
File path:
hudi-flink/src/test/java/org/apache/hudi/sink/TestWriteCopyOnWrite.java
##########
@@ -380,12 +380,12 @@ public void testUpsertWithDelete() throws Exception {
@Test
public void testInsertWithMiniBatches() throws Exception {
// reset the config option
- conf.setDouble(FlinkOptions.WRITE_BATCH_SIZE, 0.0006); // 630 bytes batch
size
+ conf.setDouble(FlinkOptions.WRITE_BATCH_SIZE, 0.00075); // 786 bytes batch
size
Review comment:
Okey, got your idea, can we keep the type as long if the data type is
timestamp without time zone (or UTC timezone), we can keep the compatibility
and the long has less memory footprint as you have tested.
This is important because Hoodie writer handle buffer the records, so the
less memory the better.
--
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.
For queries about this service, please contact Infrastructure at:
[email protected]
