Hello Tejun On Tue, 15 Oct 2019 07:37:31 -0700 Tejun Heo wrote: > Hello, Hillf. > > Do you have a test case which can demonstrate the problem you're > seeing in the existing code? I dont have such a test case because I see no problem in the current bw measurings except for the difficulties. For example, wb-A's bw is measured to be 66MB/s if wb-B joins it dispatching IO, or 96MB/s if wb-C also joins them, in assumption it is a simple case. It may be too difficult to be feasible, I am afraid, to get wb-A's bw under the workloads in data centers without wb-non-A's churnings. Thanks Hillf
