This would be reduced to a great extent if request throttling is brought into heketi. Work on that is in progress [1], [2] however it won't be done by CNS 3.10 time frame. Also, Openshift(storage controller, provisioner) does not keep track of its request beyond the time out period. It needs some changes too. I suggest we move this out of CNS 3.10 [1] https://github.com/heketi/heketi/pull/1131 [2] https://github.com/heketi/heketi/pull/1128