Abhishek Chennaka has posted comments on this change. ( http://gerrit.cloudera.org:8080/24578 )
Change subject: Add diff-scan option to return unobservable rows ...................................................................... Patch Set 7: @Zoltan: You're right on both counts. First point: no, diff scan does not report values 6 and 7 as separate outputs. The iterator collapses all mutations for a row within the range and emits a single row reflecting its state at snap_end. In your first example the diff scan reports "a" as one row with IS_DELETED=true (via include_deleted_rows). Second point: yes, exactly. The scenario this commit addresses is your second one — a row deleted before snap_start (because it was migrated out and the delete was propagated) that goes through a full lifecycle inside the diff-scan window. Today no diff scan surfaces such a row at all, and any stale copy in the second tier is never reconciled. The new INCLUDE_UNOBSERVABLE mode surfaces the row once with IS_DELETED=true. Let me update the commit message to clarify on the CDC framing along with the Jira. -- To view, visit http://gerrit.cloudera.org:8080/24578 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: kudu Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I0d0727fb2b89c727c669ec218e3bb8594336359a Gerrit-Change-Number: 24578 Gerrit-PatchSet: 7 Gerrit-Owner: Abhishek Chennaka <[email protected]> Gerrit-Reviewer: Abhishek Chennaka <[email protected]> Gerrit-Reviewer: Alexey Serbin <[email protected]> Gerrit-Reviewer: Ashwani Raina <[email protected]> Gerrit-Reviewer: Kudu Jenkins (120) Gerrit-Reviewer: Marton Greber <[email protected]> Gerrit-Reviewer: Michael Smith <[email protected]> Gerrit-Reviewer: Zoltan Martonka <[email protected]> Gerrit-Comment-Date: Fri, 14 Aug 2026 23:32:38 +0000 Gerrit-HasComments: No
