The current connection scheduling mechanism (RoundRobin) doesn't take into account the qpair type and assigns each new qpair to the next poll group. As a side effect there might occur a disbalance when some poll group handles more IO qpairs than others. In RDMA transport it is possible to get the qpair type before the controller creation using a private data from the rdma_cm event, this allows to schedule admin and IO qpairs in the balanced way. Change-Id: I90c368a41c4cd0f5347a83cab7511e4494f05b29 Signed-off-by: Alexey Marchuk <alexeymar@mellanox.com> Signed-off-by: Sasha Kotchubievsky <sashakot@mellanox.com> Signed-off-by: Evgenii Kochetov <evgeniik@mellanox.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/468993 Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Jim Harris <james.r.harris@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Community-CI: Broadcom SPDK FC-NVMe CI <spdk-ci.pdl@broadcom.com> |
||
---|---|---|
.. | ||
ctrlr_bdev.c | ||
ctrlr_discovery.c | ||
ctrlr.c | ||
fc_ls.c | ||
fc.c | ||
Makefile | ||
nvmf_fc.h | ||
nvmf_internal.h | ||
nvmf_rpc.c | ||
nvmf.c | ||
rdma.c | ||
subsystem.c | ||
tcp.c | ||
transport.c | ||
transport.h |