Discover and connect to the storage from the host

Before making definitions of each host in ThinkSystem System Manager, you must discover the target controller ports from the host, and then establish NVMe connections.

1. Discover available subsystems on the NVMe-oF target for all paths using the following command:

   nvme discover -t rdma -a target_ip_address

   In this command, target_ip_address is the IP address of the target port.

   Note: The nvme discover command discovers all controller ports in the subsystem, regardless of host access.

   # nvme discover -t rdma -a 192.168.1.77
   Discovery Log Number of Records 2, Generation counter 0
   =====Discovery Log Entry 0======
   trtype:  rdma
   adrfam:  ipv4
   subtype: nvme subsystem
   treq:    not specified
   portid:  0
   trsvcid: 4420
   subnqn:  nqn.1992-08.com.netapp:5700.600a098000a527a7000000005ab3af94
   traddr:  192.168.1.77
   rdma_prtype: roce
   rdma_qptype: connected
   rdma_cms:    rdma-cm
   rdma_pkey: 0x0000
   =====Discovery Log Entry 1======
   trtype:  rdma
   adrfam:  ipv4
   subtype: nvme subsystem
   treq:    not specified
   portid:  1
   trsvcid: 4420
   subnqn:  nqn.1992-08.com.netapp:5700.600a098000a527a7000000005ab3af94
   traddr:  192.168.2.77
   rdma_prtype: roce
   rdma_qptype: connected
   rdma_cms:    rdma-cm
   rdma_pkey: 0x0000

2. Repeat step 1 for any other connections.
3. Connect to the discovered subsystem on the first path using the command: nvme connect -t rdma -n discovered_sub_nqn -a target_ip_address -Q queue_depth_setting -l controller_loss_timeout_period
   Note: The nvme connect -t rdma -n discovered_sub_nqn -a target_ip_address -Q queue_depth_setting -l controller_loss_timeout_period command does not persist through reboot. The NVMe connect command will need to executed after each reboot to re-establish the NVMe connections.

   Important: Connections are not established for any discovered port inaccessible by the host.

   Important: If you specify a port number using this command, the connection fails. The default port is the only port set up for connections.

   Important: The recommended queue depth setting is 1024. Override the default setting of 128 with 1024 using the -Q 1024 command line option, as shown in the following example.

   Important: The recommended controller loss timeout period in seconds is 60 minutes (3600 seconds). Override the default setting of 600 seconds with 3600 seconds using the -l 3600 command line option, as shown in the following example.

   # nvme connect -t rdma -a 192.168.1.77 -n nqn.1992-08.com.netapp:5700.
   600a098000a527a7000000005ab3af94 -Q 1024 -l 3600
   # nvme connect -t rdma -a 192.168.2.77 -n nqn.1992-08.com.netapp:5700.
   600a098000a527a7000000005ab3af94 -Q 1024 -l 3600

4. Repeat step 3 to connect the discovered subsystem on the second path.
5. Auto connect after system reboot setup.
   1. Create file nvmf-autoconnect.service under /usr/lib/systemd/system, if the file doesn't exist.
   2. Fulfill following service content as below:

      [Unit]
      Description=Connect NVMe-oF subsystems automatically during boot
      ConditionPathExists=/etc/nvme/discovery.conf
      After=network.target
      Before=remote-fs-pre.target
      
      [Service]
      Type=oneshot
      ExecStart=/usr/sbin/nvme connect-all
      
      [Install]
      WantedBy=default.target