Configure Proxmox VE 9.x for NVMe-oF with ONTAP storage

12/16/2025 Contributors

The Proxmox VE 9.x host supports the NVMe over Fibre Channel (NVMe/FC) and NVMe over TCP (NVMe/TCP) protocols with Asymmetric Namespace Access (ANA). ANA provides multipathing functionality equivalent to asymmetric logical unit access (ALUA) in iSCSI and FCP environments.

Learn how to configure NVMe over Fabrics (NVMe-oF) hosts for Proxmox VE 9.x. For more support and feature information, see ONTAP support and features.

NVMe-oF with Proxmox VE 9.x has the following known limitation:

The SAN boot configuration for NVMe-FC is not supported.

Step 1: Install Proxmox VE and NVMe software and verify your configuration

To configure your host for NVMe-oF you need to install the host and NVMe software packages, enable multipathing, and verify your host NQN configuration.

Steps

Install Proxmox VE 9.x on the server. After the installation is complete, verify that you are running the required Proxmox VE 9.x kernel:
```
uname -r
```
Example Proxmox VE 9.x kernel version:
```
6.14.8-2-pve
```
Install the nvme-cli package:
```
apt list|grep nvme-cli
```
The following example shows an nvme-cli package version:
```
nvme-cli/stable,now 2.13-2 amd64
```
Install the libnvme package:
```
apt list|grep libnvme
```
The following example shows an libnvme package version:
```
libnvme-dev/stable 1.13-2 amd64
```
On the host, check the hostnqn string at /etc/nvme/hostnqn:
```
cat /etc/nvme/hostnqn
```
The following example shows a hostnqn value:
```
nqn.2014-08.org.nvmexpress:uuid:39333550-3333-4753-4844-32594d4a524c
```

On the ONTAP system, verify that the hostnqn string matches the hostnqn string for the corresponding subsystem on the ONTAP array:

::> vserver nvme subsystem host show -vserver vs_proxmox_FC_NVMeFC

Show example

Vserver Subsystem Priority  Host NQN
------- --------- --------  -----------------------------------------------
vs_proxmox_FC_NVMeFC
        sub_176
                  regular   nqn.2014-08.org.nvmexpress:uuid:39333550-3333-4753-4844-32594d4a4834
                  regular   nqn.2014-08.org.nvmexpress:uuid:39333550-3333-4753-4844-32594d4a524c
2 entries were displayed

If the hostnqn strings do not match, use the vserver modify command to update the hostnqn string on your corresponding ONTAP storage system subsystem to match the hostnqn string from /etc/nvme/hostnqn on the host.

Step 2: Configure NVMe/FC and NVMe/TCP

Configure NVMe/FC with Broadcom/Emulex or Marvell/QLogic adapters, or configure NVMe/TCP using manual discovery and connect operations.

NVMe/FC - Broadcom/Emulex

Configure NVMe/FC for a Broadcom/Emulex adapter.

Verify that you are using the supported adapter model:
1. Display the model names:
  cat /sys/class/scsi_host/host*/modelname
  You should see the following output:
  SN1700E2P SN1700E2P
2. Show the model descriptions:
  cat /sys/class/scsi_host/host*/modeldesc
  You should see output similar to the following example:
  HPE SN1700E 64Gb 2p FC HBA HPE SN1700E 64Gb 2p FC HBA
Verify that you are using the recommended Broadcom lpfc firmware and inbox driver:
1. Display the firmware version:
  cat /sys/class/scsi_host/host*/fwrev
  The command returns the firmware versions:
  14.4.473.14, sli-4:6:d 14.4.473.14, sli-4:6:d
2. Display the inbox driver version:
  cat /sys/module/lpfc/version
  The following example shows a driver version:
  0:14.4.0.7
For the current list of supported adapter driver and firmware versions, see the Interoperability Matrix Tool.

Verify that lpfc_enable_fc4_type is set to 3:

cat /sys/module/lpfc/parameters/lpfc_enable_fc4_type

Verify that you can view your initiator ports:
```
cat /sys/class/fc_host/host*/port_name
```
You should see output similar to:
```
0x10005ced8c531948
0x10005ced8c531949
```
Verify that your initiator ports are online:
```
cat /sys/class/fc_host/host*/port_state
```
You should see the following output:
```
Online
Online
```

Verify that the NVMe/FC initiator ports are enabled and that the target ports are visible:

cat /sys/class/scsi_host/host*/nvme_info

Show example

NVME Initiator Enabled
XRI Dist lpfc0 Total 6144 IO 5894 ELS 250
NVME LPORT lpfc0 WWPN x10005ced8c531948 WWNN x20005ced8c531948 DID x082400 ONLINE
NVME RPORT       WWPN x200ed039eac79573 WWNN x200dd039eac79573 DID x060902 TARGET DISCSRVC ONLINE
NVME RPORT       WWPN x2001d039eac79573 WWNN x2000d039eac79573 DID x060904 TARGET DISCSRVC ONLINE

NVME Statistics
LS: Xmt 0000000034 Cmpl 0000000034 Abort 00000000
LS XMIT: Err 00000000  CMPL: xb 00000000 Err 00000000
Total FCP Cmpl 0000000000142cfb Issue 0000000000142cfc OutIO 0000000000000001
        abort 00000005 noxri 00000000 nondlp 00000000 qdepth 00000000 wqerr 00000000 err 00000000
FCP CMPL: xb 00000005 Err 00000005

NVME Initiator Enabled
XRI Dist lpfc1 Total 6144 IO 5894 ELS 250
NVME LPORT lpfc1 WWPN x10005ced8c531949 WWNN x20005ced8c531949 DID x082500 ONLINE
NVME RPORT       WWPN x2010d039eac79573 WWNN x200dd039eac79573 DID x062902 TARGET DISCSRVC ONLINE
NVME RPORT       WWPN x2007d039eac79573 WWNN x2000d039eac79573 DID x062904 TARGET DISCSRVC ONLINE

NVME Statistics
LS: Xmt 0000000034 Cmpl 0000000034 Abort 00000000
LS XMIT: Err 00000000  CMPL: xb 00000000 Err 00000000
Total FCP Cmpl 00000000000d39f1 Issue 00000000000d39f2 OutIO 0000000000000001
        abort 00000005 noxri 00000000 nondlp 00000000 qdepth 00000000 wqerr 00000000 err 00000000
FCP CMPL: xb 00000005 Err 00000005

NVMe/FC - Marvell/QLogic

Configure NVMe/FC for a Marvell/QLogic adapter.

Verify that you are using the supported adapter driver and firmware versions:
```
cat /sys/class/fc_host/host*/symbolic_name
```
The following example shows driver and firmware versions:
```
SN1700Q FW:v9.15.05 DVR:v10.02.09.400-k
SN1700Q FW:v9.15.05 DVR:v10.02.09.400-k
```
Verify that ql2xnvmeenable is set. This enables the Marvell adapter to function as an NVMe/FC initiator:
```
cat /sys/module/qla2xxx/parameters/ql2xnvmeenable
```
The expected output is 1.

NVMe/TCP

The NVMe/TCP protocol doesn't support the auto-connect operation. Instead, you can discover the NVMe/TCP subsystems and namespaces by performing the NVMe/TCP connect or connect-all operations manually.

Check that the initiator port can get the discovery log page data across the supported NVMe/TCP LIFs:

nvme discover -t tcp -w host-traddr -a traddr

Show example

nvme discover -t tcp -w 192.168.165.72 -a 192.168.165.51
Discovery Log Number of Records 4, Generation counter 47
=====Discovery Log Entry 0======
trtype:  tcp
adrfam:  ipv4
subtype: current discovery subsystem
treq:    not specified
portid:  3
trsvcid: 8009
subnqn:  nqn.1992-08.com.netapp:sn.c770be5d934811f0b624d039eac809ba:discovery
traddr:  192.168.165.51
eflags:  explicit discovery connections, duplicate discovery information
sectype: none
=====Discovery Log Entry 1======
trtype:  tcp
adrfam:  ipv4
subtype: current discovery subsystem
treq:    not specified
portid:  1
trsvcid: 8009
subnqn:  nqn.1992-08.com.netapp:sn.c770be5d934811f0b624d039eac809ba:discovery
traddr:  192.168.166.50
eflags:  explicit discovery connections, duplicate discovery information
sectype: none
=====Discovery Log Entry 2======
trtype:  tcp
adrfam:  ipv4
subtype: nvme subsystem
treq:    not specified
portid:  3
trsvcid: 4420
subnqn:  nqn.1992-08.com.netapp:sn.c770be5d934811f0b624d039eac809ba:subsystem.sub_176
traddr:  192.168.165.51
eflags:  none
sectype: none
=====Discovery Log Entry 3======
trtype:  tcp
adrfam:  ipv4
subtype: nvme subsystem
treq:    not specified
portid:  1
trsvcid: 4420
subnqn:  nqn.1992-08.com.netapp:sn.c770be5d934811f0b624d039eac809ba:subsystem.sub_176
traddr:  192.168.166.50
eflags:  none
sectype: none

Run the nvme connect-all command across all the supported NVMe/TCP initiator-target LIFs across the nodes:

nvme connect-all -t tcp -w host-traddr -a traddr

Show example

nvme connect-all -t tcp -w 192.168.166.73 -a 192.168.166.50
nvme connect-all -t tcp -w 192.168.166.73 -a 192.168.166.51
nvme connect-all -t tcp -w 192.168.165.73 -a 192.168.165.50
nvme connect-all -t tcp -w 192.168.165.73 -a 192.168.165.51

The setting for the NVMe/TCP ctrl_loss_tmo timeout is automatically set to "off". As a result:

There are no limits on the number of retries (indefinite retry).
You don't need to manually configure a specific ctrl_loss_tmo timeout duration when using the nvme connect or nvme connect-all commands (option -l ).
The NVMe/TCP controllers don't experience timeouts in the event of a path failure and remain connected indefinitely.

Step 3: Optionally, enable 1MB I/O for NVMe/FC

ONTAP reports a Max Data Transfer Size (MDTS) of 8 in the Identify Controller data. This means the maximum I/O request size can be up to 1MB. To issue I/O requests of size 1MB for a Broadcom NVMe/FC host, you should increase the lpfc value of the lpfc_sg_seg_cnt parameter to 256 from the default value of 64.

These steps don't apply to Qlogic NVMe/FC hosts.

Steps

Set the lpfc_sg_seg_cnt parameter to 256:
```
cat /etc/modprobe.d/lpfc.conf
```
You should see an output similar to the following example:
```
options lpfc lpfc_sg_seg_cnt=256
```
Run the update-initramfs command, and reboot the host.

Verify that the value for lpfc_sg_seg_cnt is 256:

cat /sys/module/lpfc/parameters/lpfc_sg_seg_cnt

Step 4: Verify NVMe boot services

The nvmefc-boot-connections.service and nvmf-autoconnect.service boot services included in the NVMe/FC nvme-cli package are automatically enabled when the system boots.

After booting completes, verify that the nvmefc-boot-connections.service and nvmf-autoconnect.service boot services are enabled.

Steps

Verify that nvmf-autoconnect.service is enabled:

systemctl status nvmf-autoconnect.service

Show example output

○ nvmf-autoconnect.service - Connect NVMe-oF subsystems automatically during boot
     Loaded: loaded (/usr/lib/systemd/system/nvmf-autoconnect.service; enabled; preset: enabled)
     Active: inactive (dead) since Thu 2025-10-16 18:00:39 IST; 6 days ago
 Invocation: e146e0b2c339432aad6e0555a528872c
    Process: 1787 ExecStart=/usr/sbin/nvme connect-all --context=autoconnect (code=exited, status=0/SUCCESS)
   Main PID: 1787 (code=exited, status=0/SUCCESS)
   Mem peak: 2.4M
        CPU: 12ms
Oct 16 18:00:39 HPE-DL365-14-176 systemd[1]: Starting nvmf-autoconnect.service - Connect NVMe-oF subsystems automatically during boot...
Oct 16 18:00:39 HPE-DL365-14-176 systemd[1]: nvmf-autoconnect.service: Deactivated successfully.
Oct 16 18:00:39 HPE-DL365-14-176 systemd[1]: Finished nvmf-autoconnect.service - Connect NVMe-oF subsystems automatically during boot.

Verify that nvmefc-boot-connections.service is enabled:

systemctl status nvmefc-boot-connections.service

Show example output

nvmefc-boot-connections.service - Auto-connect to subsystems on FC-NVME devices found during boot
     Loaded: loaded (/usr/lib/systemd/system/nvmefc-boot-connections.service; enabled; preset: enabled)
     Active: inactive (dead) since Thu 2025-10-16 18:00:35 IST; 6 days ago
 Invocation: acf73ac1ef7a402198d6ecc4d075fab0
    Process: 1173 ExecStart=/bin/sh -c echo add > /sys/class/fc/fc_udev_device/nvme_discovery (code=exited, status=0/SUCCESS)
   Main PID: 1173 (code=exited, status=0/SUCCESS)
   Mem peak: 2.1M
        CPU: 11ms

Oct 16 18:00:35 HPE-DL365-14-176 systemd[1]: nvmefc-boot-connections.service: Deactivated successfully.
Oct 16 18:00:35 HPE-DL365-14-176 systemd[1]: Finished nvmefc-boot-connections.service - Auto-connect to subsystems on FC-NVME devices found during boot.

Step 5: Verify the multipathing configuration

Verify that the in-kernel NVMe multipath status, ANA status, and ONTAP namespaces are correct for the NVMe-oF configuration.

Steps

Verify that the in-kernel NVMe multipath is enabled:
```
cat /sys/module/nvme_core/parameters/multipath
```
You should see the following output:
```
Y
```
Verify that the appropriate NVMe-oF settings (such as, model set to NetApp ONTAP Controller and load balancing iopolicy set to round-robin) for the ONTAP namespaces show up correctly on the host:
1. Show the subsystems:
  cat /sys/class/nvme-subsystem/nvme-subsys*/model
  You should see the following output:
  NetApp ONTAP Controller NetApp ONTAP Controller
2. Display the policy:
  cat /sys/class/nvme-subsystem/nvme-subsys*/iopolicy
  You should see the following output:
  queue-depth queue-depth

Verify that the namespaces are created and correctly discovered on the host:

nvme list

Show example

Node          Generic     SN                   Model                    Namespace  Usage                      Format           FW Rev
------------- ----------- -------------------- ------------------------ ---------- -------------------------- ---------------- --------
/dev/nvme2n1  /dev/ng2n1  81PqYFYq2aVAAAAAAAAB NetApp ONTAP Controller   0x1         17.88  GB / 171.80  GB      4 KiB +  0 B   9.17.1

Verify that the controller state of each path is live and has the correct ANA status:

NVMe/FC

nvme list-subsys /dev/nvme3n9

Show example

nvme-subsys3 - NQN=nqn.1992-08.com.netapp:sn.94929fdb84eb11f0b624d039eac809ba:subsystem.sub_176
               hostnqn=nqn.2014-08.org.nvmexpress:uuid:39333550-3333-4753-4844-32594d4a524c
\
 +- nvme1 fc traddr=nn-0x200dd039eac79573:pn-0x2010d039eac79573,host_traddr=nn-0x20005ced8c531949:pn-0x10005ced8c531949 live optimized
 +- nvme3 fc traddr=nn-0x200dd039eac79573:pn-0x200ed039eac79573,host_traddr=nn-0x20005ced8c531948:pn-0x10005ced8c531948 live optimized
+- nvme5 fc traddr=nn-0x200dd039eac79573:pn-0x200fd039eac79573,host_traddr=nn-0x20005ced8c531949:pn-0x10005ced8c531949 live non-optimized
 +- nvme7 fc traddr=nn-0x200dd039eac79573:pn-0x2011d039eac79573,host_traddr=nn-0x20005ced8c531948:pn-0x10005ced8c531948 live non-optimized

NVMe/TCP

nvme list-subsys /dev/nvme2n3

Show example

nvme-subsys2 - NQN=nqn.1992-08.com.netapp:sn.c770be5d934811f0b624d039eac809ba:subsystem.sub_176
               hostnqn=nqn.2014-08.org.nvmexpress:uuid:39333550-3333-4753-4844-32594d4a524c
\
 +- nvme2 tcp traddr=192.168.166.50,trsvcid=4420,host_traddr=192.168.166.73,src_addr=192.168.166.73 live optimized
 +- nvme4 tcp traddr=192.168.165.51,trsvcid=4420,host_traddr=192.168.165.73,src_addr=192.168.165.73 live optimized
+- nvme6 tcp traddr=192.168.166.51,trsvcid=4420,host_traddr=192.168.166.73,src_addr=192.168.166.73 live non-optimized
 +- nvme8 tcp traddr=192.168.165.50,trsvcid=4420,host_traddr=192.168.165.73,src_addr=192.168.165.73 live non-optimized

Verify that the NetApp plug-in displays the correct values for each ONTAP namespace device:

Column

nvme netapp ontapdevices -o column

Show example

Device        Vserver               Namespace Path
------------- --------------------- ------------------------------
/dev/nvme2n9  vs_proxmox_FC_NVMeFC  /vol/vol_180_data_nvmefc4/ns

NSID       UUID                            Size
---- ------------------------------------  --------
1    e3d3d544-de8b-4787-93af-bfec7769e909  32.21GB

JSON

nvme netapp ontapdevices -o json

Show example

{
      "Device":"/dev/nvme2n9",
      "Vserver":"vs_proxmox_FC_NVMeFC",
      "Subsystem":"sub_176",
      "Namespace_Path":"/vol/vol_180_data_nvmefc4/ns",
      "NSID":9,
      "UUID":"e3d3d544-de8b-4787-93af-bfec7769e909",
      "LBA_Size":4096,
      "Namespace_Size":32212254720,
      "UsedBytes":67899392,
      "Version":"9.17.1"
    }
  ]

Step 6: Review the known issues

There are no known issues.

Configure Proxmox VE 9.x for NVMe-oF with ONTAP storage

Creating your file...

Step 1: Install Proxmox VE and NVMe software and verify your configuration

Step 2: Configure NVMe/FC and NVMe/TCP

Step 3: Optionally, enable 1MB I/O for NVMe/FC

Step 4: Verify NVMe boot services

Step 5: Verify the multipathing configuration

Step 6: Review the known issues