Mike Percy has posted comments on this change. Change subject: docs: clarify steps for removing master from multi-master deployment ......................................................................
Patch Set 2: (1 comment) http://gerrit.cloudera.org:8080/#/c/8032/2/docs/administration.adoc File docs/administration.adoc: PS2, Line 382: WARNING: dropping the number of masters below the number of masters currently needed for a Raft : majority can incur data loss. > Please double check this with Mike. You are technically correct (the best kind of correct) but there are nuances we might consider explaining here. Data loss is kind of a funny concept in this case because we're really talking about *metadata* loss, which can actually be catastrophic: for example it can cause the master to end up dropping whole tables, I think, depending on how far behind the remaining node was, like if it went offline before you created a new table or partition. However, I'm not sure whether we've ever tested this scenario and I'm pretty confident that we don't have an *automated* test for it either. This is particularly scary if you had one master that was down or partitioned for a long time, and you end up removing the other two, and then this stale remaining master comes back online and is all we have left. In general, if you remove more than a majority at once and then do some kind of manual repair, we don't give you any durability guarantees at all. -- To view, visit http://gerrit.cloudera.org:8080/8032 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: comment Gerrit-Change-Id: I4196dbb2f8a185e868a6906c7cf917d79c404c0d Gerrit-PatchSet: 2 Gerrit-Project: kudu Gerrit-Branch: master Gerrit-Owner: Andrew Wong <aw...@cloudera.com> Gerrit-Reviewer: Adar Dembo <a...@cloudera.com> Gerrit-Reviewer: Alexey Serbin <aser...@cloudera.com> Gerrit-Reviewer: Andrew Wong <aw...@cloudera.com> Gerrit-Reviewer: Kudu Jenkins Gerrit-Reviewer: Mike Percy <mpe...@apache.org> Gerrit-HasComments: Yes