From mboxrd@z Thu Jan 1 00:00:00 1970 From: kbusch@kernel.org (Keith Busch) Date: Thu, 23 May 2019 07:39:46 -0600 Subject: [PATCH 9/9] nvme: Retrieve namespaces during list-subsys cmd In-Reply-To: <0c677aa1-aa60-dddd-4a3f-a339545fee63@mellanox.com> References: <1558543193-24752-1-git-send-email-maxg@mellanox.com> <1558543193-24752-10-git-send-email-maxg@mellanox.com> <20190522224955.GC5857@localhost.localdomain> <0c677aa1-aa60-dddd-4a3f-a339545fee63@mellanox.com> Message-ID: <20190523133946.GE14049@localhost.localdomain> On Thu, May 23, 2019@01:01:54AM -0700, Max Gurtovoy wrote: > > On 5/23/2019 1:49 AM, Keith Busch wrote: > > On Wed, May 22, 2019@07:39:53PM +0300, Max Gurtovoy wrote: > >> Add an association between subsystems/ctrls/namespaces using > >> "nvme list-subsys" command. Now this command will show the following: > >> > >> nvme-subsys4 - NQN=testsubsystem_0 > >> \ > >> +- nvme4 rdma traddr=12.212.99.85 trsvcid=4420 live > >> \ > >> +- nvme4n1 > >> +- nvme4n2 > >> +- nvme4n3 > >> +- nvme5 rdma traddr=12.212.99.85 trsvcid=4420 live > >> \ > >> +- nvme5n1 > >> +- nvme5n2 > >> +- nvme5n3 > >> > >> Instead of: > >> ---------------- > >> nvme-subsys4 - NQN=testsubsystem_0 > >> \ > >> +- nvme4 rdma traddr=12.212.99.85 trsvcid=4420 > >> +- nvme5 rdma traddr=12.212.99.85 trsvcid=4420 > > This is a terrific start. Your output indicates you are not using > > nvme native multipathing, though. Could you retry this command with > > that enabled (I'll try it tomorrow as well)? > > Yup I'll configure the native multipath. But what is expected ? > > should we print nvme4c1n1/nvme4c1n2/nvme4c1n3 for nvme4 > > and nvme4c2n1/nvme4c2n2/nvme4c2n3 for nvme5 ? > > if so, I guess the information on the "IO-ble" devices > nvme4n1/nvme4n2/nvme4n3 will not be printed. > > something like: > > nvme-subsys4 - NQN=testsubsystem_0 > \ > ?+- nvme4 rdma traddr=12.212.99.85 trsvcid=4420 live > ?\ > ? +- nvme4c1n1 > ? +- nvme4c1n2 > ? +- nvme4c1n3 > ?+- nvme5 rdma traddr=12.212.99.85 trsvcid=4420 live > ?\ > ? +- nvme4c2n1 > ? +- nvme4c2n2 > ? +- nvme4c2n3 > > > if this if fine I'll send a V2 with a small fix. I think we'd actually prefer not seeing the hidden block devices, but instead use the device name from the "head": nvme4n1, nvme4n2, etc...