freemandealer commented on code in PR #30633:
URL: https://github.com/apache/doris/pull/30633#discussion_r1473751411
##########
be/src/olap/rowset/beta_rowset_writer.cpp:
##########
@@ -552,6 +564,13 @@ Status BetaRowsetWriter::_close_file_writers() {
} else {
RETURN_NOT_OK_STATUS_WITH_WARN(_wait_flying_segcompaction(),
"segcompaction failed when build
new rowset");
+ // compact all remaining segments, and since we can't submit a
segcompaction
+ // when `_is_doing_segcompaction` is true, wait for the flying
segcompaction
+ // to finish first, before doing the check
+ RETURN_NOT_OK_STATUS_WITH_WARN(_segcompaction_if_necessary(true),
Review Comment:
As the 'compact remaining' operation is critical for the close wait latency,
queueing in the segcompaction task queue is the last thing we want. So maybe it
should have a higher priority than those normal & best-effort segcompaction
task?
##########
be/src/common/config.cpp:
##########
@@ -942,6 +942,9 @@ DEFINE_Bool(enable_segcompaction, "false");
// Max number of segments allowed in a single segcompaction task.
DEFINE_Int32(segcompaction_batch_size, "10");
+// the score threshold for last batch
+DEFINE_Int32(segcompaction_last_batch_score, "1000");
Review Comment:
Since the adaptive configuration is unavailable now, would you please
describe the meaning more clearly and add a detailed entry in the docs?
--
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]
