Details
-
Bug
-
Status: Resolved
-
Resolution: Done
-
Beryllium
-
None
-
None
-
Operating System: All
Platform: All
-
5656
Description
In a 5 node cluster when an entity-ownership shard leader and follower go down simultaneously only the previous leader's candidates are removed from the data when a new leader is elected. This leads to the follower node remaining as candidate even though it is down.
Steps to reproduce,
1. Bring up a 5 node cluster
2. Point an openflow device to all five nodes in the cluster. This should result in a candidate registering for each node in the cluster.
3. Now let's say member-2 is the entity-ownership shard leader
4. Kill the controller process for both member-1 and member-2.
5. After ~10 seconds leadership of the entity-ownership shard should change. At this time check the entity-ownership data using RestConf. You will see that member-1 which still be a candidate for the openflow entity even though member-2 is not.