Andrew Wong has posted comments on this change. ( http://gerrit.cloudera.org:8080/15145 )
Change subject: KUDU-1625: background op to GC ancient, fully deleted rowsets ...................................................................... Patch Set 10: (1 comment) http://gerrit.cloudera.org:8080/#/c/15145/10/src/kudu/tablet/tablet.cc File src/kudu/tablet/tablet.cc: http://gerrit.cloudera.org:8080/#/c/15145/10/src/kudu/tablet/tablet.cc@1481 PS10, Line 1481: if (metadata_->supports_live_row_count()) { > Hi, @Andrew Keep in mind that Kudu uses the "ancient history mark" to refer to a point before which snapshot scans will not work. But we still want to keep ancient data around so scans later than the ancient history mark will still see what the latest version of the data is, even if it was last updated several weeks ago. The goal of the DeletedRowsetGCOp is to free up storage for fully deleted rowsets that are also entirely ancient. Such rowsets aren't useful since any scan after the ancient history mark will always apply all the delete redos and return no rows. It's thus safe to entirely ignore the diskrowset entirely and get rid of them. We can't make the same case for data that hasn't been fully deleted, because the base data may still contain data that is still scannable. IIRC, without the live row count, we'd need to open each delta store to determine whether a diskrowset is fully deleted. Without doing that, I'm not sure there's a safe way to enable this without live row count support. -- To view, visit http://gerrit.cloudera.org:8080/15145 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: kudu Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I696e2a29ea52ad4e54801b495c322bc371787124 Gerrit-Change-Number: 15145 Gerrit-PatchSet: 10 Gerrit-Owner: Andrew Wong <anjuw...@g.ucla.edu> Gerrit-Reviewer: Adar Lieber-Dembo <a...@apache.org> Gerrit-Reviewer: Alexey Serbin <ale...@apache.org> Gerrit-Reviewer: Andrew Wong <anjuw...@g.ucla.edu> Gerrit-Reviewer: Attila Bukor <abu...@apache.org> Gerrit-Reviewer: Kudu Jenkins (120) Gerrit-Reviewer: Tidy Bot (241) Gerrit-Reviewer: Volodymyr Verovkin <verjov...@cloudera.com> Gerrit-Reviewer: Yingchun Lai <acelyc1112...@gmail.com> Gerrit-Comment-Date: Thu, 10 Nov 2022 19:04:22 +0000 Gerrit-HasComments: Yes