NVMe over TCP环境搭建和卸载

总结

本文讲述了如何在两个femu间搭建NVMe over TCP环境

前置知识

这里不过多解释何为NVMe over TCP，简单来说就是使用TCP协议下发NVMe命令给远端的NVMe设备。

NVMe over TCP环境搭建时，发送NVMe命令的叫做host端，接收NVMe命令的叫target端

femu启动时，默认用的是主机端的IP地址，为了两个femu之间能够通信，需要两个femu都有独立的IP，可以通过在主机端搭建虚拟网桥和虚拟网卡使两个femu拥有独立IP

1. 检查内核配置

在开始之前需要确定NVME_TCP、NVME_TARGET等内核配置是被set的

$ cat /boot/config-`uname -r` | grep NVME

# NVME Support
CONFIG_NVME_CORE=m
CONFIG_BLK_DEV_NVME=m
# CONFIG_NVME_MULTIPATH is not set
# CONFIG_NVME_HWMON is not set
CONFIG_NVME_FABRICS=m
CONFIG_NVME_FC=m
CONFIG_NVME_TCP=m
CONFIG_NVME_TARGET=m
CONFIG_NVME_TARGET_LOOP=m
CONFIG_NVME_TARGET_FC=m
# CONFIG_NVME_TARGET_FCLOOP is not set
CONFIG_NVME_TARGET_TCP=m
# end of NVME Support

2. host端和target端安装nvme-cli

在host端和target端都安装上nvme-cli

sudo apt-get install nvme-cli

3. target端配置

1. 首先在target端安装nvmet以及nvmet-tcp模块

sudo modprobe nvmet
sudo modprobe nvmet-tcp

2. 然后使用如下命令去创建一个NVMe Target subsystem和namespace

cd /sys/kernel/config/nvmet/subsystems
sudo mkdir nvme-test-target
cd nvme-test-target/
echo 1 | sudo tee -a attr_allow_any_host > /dev/null
sudo mkdir namespaces/1
cd namespaces/1

3. 找到target端的nvme设备名称，这里是/dev/nvme0n1

femu@fvm:~$ sudo nvme list
Node             SN                   Model                                    Namespace Usage                      Format           FW Rev
---------------- -------------------- ---------------------------------------- --------- -------------------------- ---------------- --------
/dev/nvme0n1     vSSD0                FEMU BlackBox-SSD Controller             1           4.29  GB /   4.29  GB    512   B +  0 B   1.0

4. 将nvme设备/dev/nvme0n1与刚刚创建的子系统进行关联

echo -n /dev/nvme0n1 |sudo tee -a device_path > /dev/null
echo 1|sudo tee -a enable > /dev/null

5. 创建端口号，配置IP和其他参数

sudo mkdir /sys/kernel/config/nvmet/ports/1
cd /sys/kernel/config/nvmet/ports/1

echo 192.168.123.4 |sudo tee -a addr_traddr > /dev/null

echo tcp|sudo tee -a addr_trtype > /dev/null
echo 4420|sudo tee -a addr_trsvcid > /dev/null
echo ipv4|sudo tee -a addr_adrfam > /dev/null

6. 最后将端口与子系统绑定

sudo ln -s /sys/kernel/config/nvmet/subsystems/nvme-test-target/ /sys/kernel/config/nvmet/ports/1/subsystems/nvme-test-target

7. 应该可以看到nvme_tcp服务被启动

femu@fvm:/sys/kernel/config/nvmet/ports/1$ dmesg |grep "nvmet_tcp"
[ 1487.076439] nvmet_tcp: enabling port 1 (192.168.123.3:4420)

target端配置到此结束

4. host端配置

1. 安装nvme、nvme-tcp模块

sudo modprobe nvme
sudo modprobe nvme-tcp

2. 在主机端使用nvme discover命令查看是否能搜索到盘

kjay@kjay:~$ sudo nvme discover -t tcp -a 192.168.123.3 -s 4420
[sudo] password for kjay:

Discovery Log Number of Records 1, Generation counter 2
=====Discovery Log Entry 0======
trtype:  tcp
adrfam:  ipv4
subtype: nvme subsystem
treq:    not specified, sq flow control disable supported
portid:  1
trsvcid: 4420
subnqn:  nvme-test-target
traddr:  192.168.123.3
sectype: none

3. 连接该nvme盘

sudo nvme connect -t tcp -n nvme-test-target -a 192.168.123.3 -s 4420 --hostnqn=nqn.2014-08.org.nvmexpress:uuid:1b4e28ba-2fa1-11d2-883f-0016d3ccabcd

4. 可以看到盘被识别为/dev/nvme1n1

kjay@kjay:~$ sudo nvme list
Node             SN                   Model                                    Namespace Usage                      Format           FW Rev
---------------- -------------------- ---------------------------------------- --------- -------------------------- ---------------- --------
/dev/nvme0n1     BTPY72440ASR128A     INTEL SSDPEKKR128G7                      1         128.04  GB / 128.04  GB    512   B +  0 B    PSF109E
/dev/nvme1n1     713dcb7e1fef4c5d     Linux                                    1           4.29  GB /   4.29  GB    512   B +  0 B   5.4.0-77

此时可以使用nvme-cli对/dev/nvme1n1下发nvme命令

5. 如何取消host端与target端的连接

sudo nvme disconnect /dev/nvme1n1 -n nvme-test-target

6. target端如何卸载nvmet模块

这个问题困扰了我蛮久，最后还是和学长一起解决了，具体细节这里不展开，有兴趣的话可以看看这个rm -r 与 rmdir 区别

使用以下命令：

sudo rm -rf /sys/kernel/config/nvmet/ports/1/subsystems/nvme-test-target
echo 0 > /sys/kernel/config/nvmet/subsystems/nvme-test-target/namespaces/1/enable
echo -n 0 > /sys/kernel/config/nvmet/subsystems/nvme-test-target/namespaces/1/device_path
sudo rmdir --ignore-fail-on-non-empty /sys/kernel/config/nvmet/subsystems/nvme-test-target/namespaces/1
sudo rmdir --ignore-fail-on-non-empty /sys/kernel/config/nvmet/subsystems/nvme-test-target
sudo rmdir --ignore-fail-on-non-empty /sys/kernel/config/nvmet/ports/1
sudo rmmod nvmet-tcp
sudo rmmod nvmet

使用rmdir而不是rm -rf来删除之前创建的目录

7. 参考文献

How to setup NVMe/TCP with NVME-oF using KVM and QEMU

NVMe-oF不只是RDMA，还有TCP