Adar Dembo has posted comments on this change. ( http://gerrit.cloudera.org:8080/13456 )
Change subject: [tablet] Support accurate count of rows ...................................................................... Patch Set 6: (1 comment) http://gerrit.cloudera.org:8080/#/c/13456/6//COMMIT_MSG Commit Message: http://gerrit.cloudera.org:8080/#/c/13456/6//COMMIT_MSG@33 PS6, Line 33: 8.For the historical tablets: : When a negative number is returned, it means it is a historical : tablet. And a historical tablet can return a positive number of : live rows after compaction and etc. > 1.BTW, do you think this is better than a superblock-level capability that' New fields are cheap, especially booleans (~2 bytes). And it might be nice to formalize a "capabilities" feature for superblocks anyway. So the question is: what's the user impact of doing a tablet-level capability thing vs. the rowset-level thing? Ultimately, the user either cares about the live row count, or doesn't care: - If the user doesn't care, it doesn't matter what we do. - If the user does care, he/she will probably want to get the feature working on all their tablets. Waiting for all tablets to recompact fully is hard/impossible (what if a rowset just doesn't need to be compacted?) so naturally the user will turn to other means, such as rereplicating all the tablets from one tserver at a time. -- To view, visit http://gerrit.cloudera.org:8080/13456 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: kudu Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I2e6378e289bb85024c29e96c2b153fc417ed6412 Gerrit-Change-Number: 13456 Gerrit-PatchSet: 6 Gerrit-Owner: helifu <hzhel...@corp.netease.com> Gerrit-Reviewer: Adar Dembo <a...@cloudera.com> Gerrit-Reviewer: Kudu Jenkins (120) Gerrit-Reviewer: helifu <hzhel...@corp.netease.com> Gerrit-Comment-Date: Mon, 03 Jun 2019 04:03:04 +0000 Gerrit-HasComments: Yes
