This is an automated email from the ASF dual-hosted git repository.
alexey pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/kudu.git
The following commit(s) were added to refs/heads/master by this push:
new 8369b3771 [tools] KUDU-3487 rebalancer checks the voter num.
8369b3771 is described below
commit 8369b3771c4ee58dcc20068626e64a212717cc65
Author: 宋家成 <[email protected]>
AuthorDate: Tue Jul 25 15:14:33 2023 +0800
[tools] KUDU-3487 rebalancer checks the voter num.
Replace rebalance for the tables whose replication factor is 1 might
stuck for a long time because the rebalancer send LeaderStepDown request
even if there is no other voter to transfer leadership.
So check the voter num before sending the request.
Change-Id: I3b3d5dfb383719f56a2d32b62d2b659b17928794
Reviewed-on: http://gerrit.cloudera.org:8080/20264
Tested-by: Kudu Jenkins
Reviewed-by: Alexey Serbin <[email protected]>
---
src/kudu/tools/tool_replica_util.cc | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/src/kudu/tools/tool_replica_util.cc
b/src/kudu/tools/tool_replica_util.cc
index f627f3297..0a0ae5a5d 100644
--- a/src/kudu/tools/tool_replica_util.cc
+++ b/src/kudu/tools/tool_replica_util.cc
@@ -514,7 +514,9 @@ Status CheckCompleteReplace(const
client::sp::shared_ptr<client::KuduClient>& cl
// replica will not be evicted prior the new non-voter replica becomes
// is promoted into voter. Demoting former leader too early might even
// delay promotion of already caught-up non-leader replica.
- if (is_all_voters &&
+ // Also, do not send LeaderStepDown request if voter num is 1 since
+ // it make no sense and will make the leader stuck for a long time.
+ if (is_all_voters && cstate.committed_config().peers().size() > 1 &&
leader_uuid == ts_uuid && leader_uuid == cstate.leader_uuid()) {
// The leader is the node we intend to remove; make it step down.
ignore_result(DoLeaderStepDown(tablet_id, leader_uuid, leader_hp,
