Details
-
Bug
-
Status: Resolved
-
Major
-
Resolution: Fixed
-
None
-
None
Description
In our observation there is cases that scm send hundreds of containers in a command to one DN for delete. DN failed to handle all theses command in a heartbeat interval(30s), so that SCM resend the command with same batch of container again.
It's time consuming for DN to open the RocksDB instance and write the delete transaction into it.