Details
-
Bug
-
Status: Resolved
-
Minor
-
Resolution: Fixed
-
2.4.13
-
None
-
Reviewed
Description
Reproduction:
1. Create a table with region replication, and ensure that the primary region and replica region are on the same RS (eg: the HBase cluster has only one RS)
create 't01', 'info', {REGION_REPLICATION => 2}
2. The first region does not split when storefile size exceed flushsize * 2, because that we get 2 regions about this table on this RS (1 primary region and 1 replica region)
I think we should ignore the replica reggion when getting the count of regions on this same regionserver.
Is my idea correct? maybe can discuss it .