Alexey Serbin has posted comments on this change. ( http://gerrit.cloudera.org:8080/14222 )
Change subject: KUDU-2069 p4: stop replication from failed servers in maintenance mode ...................................................................... Patch Set 2: (1 comment) http://gerrit.cloudera.org:8080/#/c/14222/2/src/kudu/consensus/quorum_util.cc File src/kudu/consensus/quorum_util.cc: http://gerrit.cloudera.org:8080/#/c/14222/2/src/kudu/consensus/quorum_util.cc@456 PS2, Line 456: peer.attrs().replace() > I should have been more clear on what I said. Ah, just another observation. Probably, I should have mentioned that earlier for clarity. If I'm not mistaken, only tools like 'cluster rebalancer' and 'tablet change_config move_replica' set the 'REPLACE' attribute to make replica to be moved to some other tablet server. That corresponds to ChangeConfigType of MODIFY_PEER. The automatic re-replication in master uses ADD_PEER and REMOVE_PEER correspondingly. So, if we are about to ignore the REPLACE attribute when a tablet server is in the maintenance mode, that means a running session of the rebalancer tool of an explicit replica movement would stuck until the source tablet server is back into the regular (non-maintenance mode). Maybe it's worth double-checking what we want from the operability perspective here if ignoring the REPLACE attribute for tablet servers went into the maintenance mode: * allow for accumulation of those 'REPLACE' attributes set for replicas hosted by a tablet server in the maintenance mode (and be prepared for corresponding movements once the server is back into the normal mode) * explicitly disallow setting the 'REPLACE' attribute for replicas hosted by a tablet server in the maintenance mode Another alternative would be honoring the 'REPLACE' attribute even if a tablet server is in the maintenance mode, given that the presence of the 'REPLACE' attribute means it has been set explicitly, and it was not a result of the automatic re-replication activity. -- To view, visit http://gerrit.cloudera.org:8080/14222 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: kudu Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I9a63b55011d16900c0d27eac0eb75880074204db Gerrit-Change-Number: 14222 Gerrit-PatchSet: 2 Gerrit-Owner: Andrew Wong <[email protected]> Gerrit-Reviewer: Adar Dembo <[email protected]> Gerrit-Reviewer: Alexey Serbin <[email protected]> Gerrit-Reviewer: Andrew Wong <[email protected]> Gerrit-Reviewer: Hao Hao <[email protected]> Gerrit-Reviewer: Kudu Jenkins (120) Gerrit-Comment-Date: Tue, 17 Sep 2019 00:38:48 +0000 Gerrit-HasComments: Yes
