But why I don’t see these namespaces in nvme-cli?

Will check the patch when back to the office


Sent from Mail for Windows 10

From: Verkamp, Daniel
Sent: Thursday, March 22, 2018 6:44 PM
To: Storage Performance Development Kit
Subject: Re: [SPDK] SPDK annoying messages

Hi Ernest,

Those “inactive NS” messages are skipping namespace IDs that aren’t currently active or in use; your controller probably supports namespace management, and the current NVMe examples just check each reported namespace ID (from 1 to NN as reported in the controller data) to see if they are active.

Newer controllers support a new Identify Active Namespace List command, which provides a way to retrieve this list without iterating over each namespace and calling Identify Namespace.  There’s a patch out for review that uses this new feature where available:
https://review.gerrithub.io/#/c/401957/

Thanks,
-- Daniel

From: SPDK [mailto:spdk-bounces(a)lists.01.org] On Behalf Of Ernest Zed
Sent: Thursday, March 22, 2018 4:42 AM
To: Storage Performance Development Kit <spdk(a)lists.01.org>
Subject: [SPDK] SPDK annoying messages

Hi,
I'm using Ultrastar SN200 Series PCIe SSD, the `nvme list` shows that each disk has 1 namespace, but when attaching to controller SPDK sprinkles all over with following messages:
EAL: PCI device 0000:9f:00.0 on NUMA socket 1
EAL:   probe driver: 1c58:23 spdk_nvme
Attaching to NVMe Controller at 0000:9f:00.0 [1c58:0023]
Attached to NVMe Controller at 0000:9f:00.0 [1c58:0023]
Controller HUSMR7680BDP301      (SDM00001A6F7        ): Skipping inactive NS 2
Controller HUSMR7680BDP301      (SDM00001A6F7        ): Skipping inactive NS 3
Controller HUSMR7680BDP301      (SDM00001A6F7        ): Skipping inactive NS ...
Controller HUSMR7680BDP301      (SDM00001A6F7        ): Skipping inactive NS 128

What is it? is it some SPDK bug? since when I try to erase those using `nvme delete-ns` it says there is no such a namespace.

Sincerely,
E.