Marton Greber has posted comments on this change. ( http://gerrit.cloudera.org:8080/24945 )
Change subject: KUDU-3802 add cluster_leader_skew gauge ...................................................................... Patch Set 1: (2 comments) http://gerrit.cloudera.org:8080/#/c/24945/1/src/kudu/master/ts_manager.cc File src/kudu/master/ts_manager.cc: http://gerrit.cloudera.org:8080/#/c/24945/1/src/kudu/master/ts_manager.cc@373 PS1, Line 373: int TSManager::LeaderSkew() const { LeaderSkew() correctly guards the empty case (min_count == INT_MAX ? return 0), but the adjacent ClusterSkew() (line ~356) has no such guard — on an empty/fresh cluster it returns 0 - INT_MAX, i.e. a large negative. So a scrape of a just-started master would show cluster_leader_skew=0 but cluster_replica_skew=-2147483647. Pre-existing and out of scope here, but since you're adding the sibling gauge right next to it: worth a one-line follow-up JIRA (or a TODO) to make ClusterSkew consistent? No change needed in this patch. http://gerrit.cloudera.org:8080/#/c/24945/1/src/kudu/master/ts_manager.cc@379 PS1, Line 379: if (ts->PresumedDead()) { The dead-server exclusion branch in LeaderSkew isn't covered by either new test (the master-test heartbeats keep all servers live). Low value since it mirrors ClusterSkew's pattern, but if easy, a case that lets a reporting server go presumed-dead and asserts it drops out of the skew would close the last branch. Fine to leave as-is. -- To view, visit http://gerrit.cloudera.org:8080/24945 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: kudu Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: Iec82f841934f4e310509ee1b7156040b3bba5489 Gerrit-Change-Number: 24945 Gerrit-PatchSet: 1 Gerrit-Owner: Gabriella Lotz <[email protected]> Gerrit-Reviewer: Kudu Jenkins (120) Gerrit-Reviewer: Marton Greber <[email protected]> Gerrit-Comment-Date: Fri, 25 Sep 2026 12:14:08 +0000 Gerrit-HasComments: Yes
