Yukang-Lian opened a new pull request, #66610:
URL: https://github.com/apache/doris/pull/66610

   ### What problem does this PR solve?
   
   Issue Number: None
   
   Related PR: None
   
   Problem Summary:
   
   Cloud cumulative compaction scheduling maintains an approximate score 
between metadata synchronizations. The online write path used the raw segment 
count, so a zero-segment rowset added no score even though metadata 
synchronization counts every empty rowset as one. Empty rowsets could therefore 
accumulate without increasing tablet scheduling priority.
   
   This PR clamps the online contribution of a load rowset to at least one and 
adds the same contribution for zero-segment delete-predicate rowsets, which 
bypass `CloudRowsetBuilder`. Non-empty PREPARED rowsets keep the existing 
segment-based estimate because their versions have not been assigned yet.
   
   The unit test models that PREPARED, versionless lifecycle and verifies both 
cases: a zero-segment rowset contributes one, while a three-segment rowset 
still contributes three.
   
   ### Release note
   
   Fix cloud cumulative compaction scheduling for empty rowsets.
   
   ### Check List (For Author)
   
   - Test
       - [ ] Regression test
       - [x] Unit Test
           - `./run-be-ut.sh --run --filter='CloudCompactionTest.*' -j80` (18 
tests passed)
           - `build-support/check-format.sh`
           - `git diff --check`
       - [ ] Manual test
       - [ ] No need to test or manual test.
   
   - Behavior changed:
       - [ ] No.
       - [x] Yes. Zero-segment rowsets now contribute one to the online 
approximate cumulative compaction score.
   
   - Does this need documentation?
       - [x] No.
       - [ ] Yes.
   
   ### Check List (For Reviewer who merge this PR)
   
   - [ ] Confirm the release note
   - [ ] Confirm test cases
   - [ ] Confirm document
   - [ ] Add branch pick label
   


-- 
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]

Reply via email to