Will Berkeley has posted comments on this change. ( http://gerrit.cloudera.org:8080/11869 )
Change subject: [compaction] KUDU-1400: Improve rowset compaction policy to consider merging small DRSs ...................................................................... Patch Set 3: The tablet history gc test failure looks like it was caused by this patch, but not a bug. Basically, the following happened: 0. The test is tuned to do maintenance ops constantly. 1. Rows are inserted. A flush happens while rows are still being inserted. 2. More rows are inserted. A second flush happens. 3. Wait for maintenance ops to flush some undos. 4. The clock is advanced to make the undos old enough to GC. 5. The two DRS are compacted because they are small. This GC's the undos as part of the process? Or it doesn't compact undos into redos because the AHM passed for the redos? I need to check how the undos might have done away. 6. While the compaction is working, all the undo gc ops run but can't gc the blocks of the rowsets being compacted. 7. The output rowset has no undos. The test times out waiting for its undos to be gc'd and for that to be shown in the undo op metrics. -- To view, visit http://gerrit.cloudera.org:8080/11869 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: kudu Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I7b421c6ed77d28ebab9b91a4d6fcb1e825997e6c Gerrit-Change-Number: 11869 Gerrit-PatchSet: 3 Gerrit-Owner: Will Berkeley <wdberke...@gmail.com> Gerrit-Reviewer: Adar Dembo <a...@cloudera.com> Gerrit-Reviewer: Alexey Serbin <aser...@cloudera.com> Gerrit-Reviewer: Andrew Wong <aw...@cloudera.com> Gerrit-Reviewer: Kudu Jenkins (120) Gerrit-Reviewer: Tidy Bot (241) Gerrit-Reviewer: Todd Lipcon <t...@apache.org> Gerrit-Reviewer: Will Berkeley <wdberke...@gmail.com> Gerrit-Comment-Date: Fri, 02 Nov 2018 23:23:44 +0000 Gerrit-HasComments: No