Details
-
Bug
-
Resolution: Fixed
-
Critical
-
7.1.0
-
Untriaged
-
0
-
Unknown
-
Analytics Sprint 20
Description
On CC failover, a new node will be selected as CC. Before processing the failover on the new selected CC, the current logic only waits for the remaining keep nodes to join but doesn't wait until the metadata node is active. If the new CC starts updating the partitions topology before the metadata is node is active, it will result in creating an invalid partitions topology.
The only workaround if this happens is to manually update the partitions topology using a private API.