Marton Greber has posted comments on this change. ( http://gerrit.cloudera.org:8080/25005 )
Change subject: [auto_rebalancer] enforce per-server move cap ...................................................................... Patch Set 1: Code-Review+1 (1 comment) http://gerrit.cloudera.org:8080/#/c/25005/1/src/kudu/master/auto_rebalancer.cc File src/kudu/master/auto_rebalancer.cc: http://gerrit.cloudera.org:8080/#/c/25005/1/src/kudu/master/auto_rebalancer.cc@488 PS1, Line 488: RETURN_NOT_OK(SelectReplicaToMove(move, extra_info_by_tablet_id, For a move that SelectReplicaToMove turns into a pure drop (replication_factor < num_voters -> ts_uuid_to cleared), the destination move.to is checked against the cap here but is never incremented below. So a drop can be skipped because move.to looks "full," even though executing a drop adds nothing to move.to. The check (move.to) and the increment (selected.ts_uuid_to, conditionally) are asymmetric for that path. Question: is gating drops on move.to intended as conservative behavior, or should the cap only consider the source for drop moves? It's safe either way (worst case a drop is deferred to the next round), so this is just asking you to confirm intent - a one-line comment would settle it. -- To view, visit http://gerrit.cloudera.org:8080/25005 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: kudu Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I9f6b9b200c24d215435329a37e953d75fe923c02 Gerrit-Change-Number: 25005 Gerrit-PatchSet: 1 Gerrit-Owner: Gabriella Lotz <[email protected]> Gerrit-Reviewer: Abhishek Chennaka <[email protected]> Gerrit-Reviewer: Alexey Serbin <[email protected]> Gerrit-Reviewer: Gabriella Lotz <[email protected]> Gerrit-Reviewer: Kudu Jenkins (120) Gerrit-Reviewer: Marton Greber <[email protected]> Gerrit-Reviewer: Zoltan Chovan <[email protected]> Gerrit-Comment-Date: Tue, 06 Oct 2026 15:42:07 +0000 Gerrit-HasComments: Yes
