Created on Jun 27, 2022
Introduction
This post describes the setup and configuration of SR-IOV in a ESXi 7.0 host using a NVIDIA ConnectX® -6 InfiniBand adapter card with a Native driver.
Note: Setting up a VM is out of the scope of this post.
References
How-to: Install NVIDIA Firmware Tools (MFT) on VMware ESXi 6.7/7.0.
How-to: NVIDIA ConnectX driver upgrade on VMware ESXi 6.5 and above.
How-to: Firmware update for NVIDIA ConnectX-5/6 adapter on VMware ESXi 6.5 and above.
Single Root IO Virtualization (SR-IOV)
Single Root IO Virtualization (SR-IOV) is a technology that allows a physical PCIe device to present itself multiple times through the PCIe bus. This technology enables multiple virtual instances of the device with separate resources. NVIDIA adapters are capable of exposing in ConnectX-6 adapter cards up to 128 virtual instances called Virtual Functions (VFs). These virtual functions can then be provisioned separately. Each VF can be seen as an additional device connected to the Physical Function. It shares the same resources with the Physical Function.
SR-IOV is commonly used in conjunction with an SR-IOV enabled hypervisor to provide Virtual Machines direct hardware access to network resources hence increasing its performance.
Overview
SR-IOV configuration includes the following:
Enables Virtualization (SR-IOV) in the BIOS (prerequisites)
Enables SR-IOV in the firmware
Enables SR-IOV in the MLNX_OFED Driver
Maps the Virtual Machine (VM) to the relevant port via SR-IOV
Hardware and Software Requirements
A server platform with an SR-IOV-capable motherboard BIOS
Installer Privileges: The installation requires administrator privileges on the target machine
Prerequisites
To set up an SR-IOV environment, the following is required:
Make sure that SR-IOV is enabled in the BIOS of the specific server. Each server has different BIOS configuration options for virtualization. See as sample How-to Set Dell PowerEdge R730 BIOS parameters to support SR-IOV for BIOS configuration examples.
Install NVIDIA Firmware Tools (MFT) on ESXi server, refer to How-to: Install NVIDIA Firmware Tools (MFT) on VMware ESXi 6.7/7.0.
Make sure to have the last supported firmware version.
Refer to NVIDIA ConnectX® Ethernet Driver for VMware® ESXi Server and How-to: Firmware update for NVIDIA ConnectX-5/6 adapter on VMware ESXi 6.5 and above.
Setting Up SR-IOV
Enable SR-IOV in the BIOS
Each server has a different BIOS configuration options for virtualization.
The figures used in this section are for illustration purposes only.
For further information, please refer to the appropriate BIOS User Manual.
Enable "SR-IOV" in the system BIOS.
Enable "Intel Virtualization Technology".
Enable SR-IOV on the Firmware
Enable SSH Access to ESXi server.
Log into the ESXi vSphere Command-Line Interface with root permissions.
Enter Maintenance Mode the ESXi host.
Run MFT and check the status.
ESXi Console
# /opt/mellanox/bin/mst start Module mst is already loaded # /opt/mellanox/bin/mst status -vv PCI devices: ------------ DEVICE_TYPE MST PCI RDMA NET NUMA ConnectX6(rev:0) mt4123_pciconf1 37:00.0 ConnectX6(rev:0) mt4123_pciconf1.1 37:00.1
Query the status of the device.
ESXi Console
# /opt/mellanox/bin/mlxconfig -d mt4123_pciconf1 q Device #1: ---------- Device type: ConnectX6 Name: MCX654106A-HCA_Ax Description: ConnectX-6 VPI adapter card; HDR IB (200Gb/s) and 200GbE; dual-port QSFP56; Socket Direct 2x PCIe3.0 x16; tall bracket; ROHS R6 Device: mt4123_pciconf1 Configurations: Next Boot ... NUM_OF_VFS 0 SRIOV_EN False(0) ...
Set the link type of one or both ports to InfiniBand (default).
ESXi Console
# /opt/mellanox/bin/mlxconfig -d mt4123_pciconf1 set LINK_TYPE_P1=1 Device #1: ---------- Device type: ConnectX6 Name: MCX654106A-HCA_Ax Description: ConnectX-6 VPI adapter card; HDR IB (200Gb/s) and 200GbE; dual-port QSFP56; Socket Direct 2x PCIe3.0 x16; tall bracket; ROHS R6 Device: mt4123_pciconf1 Configurations: Next Boot New LINK_TYPE_P1 ETH(2) IB(1) Apply new Configuration? (y/n) [n] : y Applying... Done! # /opt/mellanox/bin/mlxconfig -d mt4123_pciconf1 set LINK_TYPE_P2=1 Device #1: ---------- Device type: ConnectX6 Name: MCX654106A-HCA_Ax Description: ConnectX-6 VPI adapter card; HDR IB (200Gb/s) and 200GbE; dual-port QSFP56; Socket Direct 2x PCIe3.0 x16; tall bracket; ROHS R6 Device: mt4123_pciconf1 Configurations: Next Boot New LINK_TYPE_P2 ETH(2) IB(1) Apply new Configuration? (y/n) [n] : y Applying... Done! -I- Please reboot machine to load new configurations.
WarningOne InfiniBand port per subnet must be dedicated to running the Subnet Manager (SM). Since the SM can only run on PFs, that port must be passthrough to a VM.
Enable Ethernet PCI subclass override.
ESXi Console
# /opt/mellanox/bin/mlxconfig -d mt4123_pciconf1 set ADVANCED_PCI_SETTINGS=1 # /opt/mellanox/bin/mlxconfig -d mt4123_pciconf1 set FORCE_ETH_PCI_SUBCLASS=1
Reboot the server.
Enable SR-IOV and set the desired number of Virtual Functions (VFs).
SRIOV_EN=1
NUM_OF_VFS=16 ; This is an example with eight VFs per port.
ESXi Console
# /opt/mellanox/bin/mlxconfig -d mt4123_pciconf1 s SRIOV_EN=1 NUM_OF_VFS=16 Device #1: ---------- Device type: ConnectX6 Name: MCX654106A-HCA_Ax Description: ConnectX-6 VPI adapter card; HDR IB (200Gb/s) and 200GbE; dual-port QSFP56; Socket Direct 2x PCIe3.0 x16; tall bracket; ROHS R6 Device: mt4123_pciconf1 Configurations: Next Boot New SRIOV_EN False(0) True(1) NUM_OF_VFS 0 16 Apply new Configuration? (y/n) [n] : y Applying... Done! -I- Please reboot machine to load new configurations.
NoteNote: mlxconfig must be performed for each PCI device (adapter). In parallel, in the driver the configuration is per module, which means that it will be applicable for all adapters installed on the server.
Reboot the server.
NoteNote: At this point, the VFs are not seen when using "lspci". Only when SR-IOV is enabled on the driver will you be able to see them.
Check if SR-IOV is enabled in the firmware.
ESXi Console
# /opt/mellanox/bin/mlxconfig -d mt4123_pciconf1 q Device #1: ---------- Device type: ConnectX6 Name: MCX654106A-HCA_Ax Description: ConnectX-6 VPI adapter card; HDR IB (200Gb/s) and 200GbE; dual-port QSFP56; Socket Direct 2x PCIe3.0 x16; tall bracket; ROHS R6 Device: mt4123_pciconf1 Configurations: Current ... NUM_OF_VFS 16 SRIOV_EN True(1) ...
Enable SR-IOV on the Driver
Get the module parameter list as follows:
ESXi Console
# esxcli system module parameters list -m nmlx5_core Name Type Value Description ... max_vfs array of uint Number of PCI VFs to initialize Values : Array of 'uint' of range 0-128, May be limited by device, 0 - disabled Default: 0 ...
Enable SR-IOV in the driver and set the max_vfs module parameter.
ESXi Console
# esxcli system module parameters set -m nmlx5_core -p "max_vfs=16,16"
Or, if your card second NVIDIA card.
Sample:ESXi Console
# esxcli system module parameters set -m nmlx5_core -p "max_vfs=0,0,16,16"
WarningNote 1: Allow at least one more VF to be configured on the firmware (num_of_vfs) than on the driver. In our example we had eight VFs configured on the firmware while four is configured on the driver (max_vfs).
Note 2: mlxconfig must be performed for each PCI device (adapter). In parallel, in the driver the configuration is per module, which means that it will be applicable for all adapters installed on the server.
Note 3: Changing the number of VFs is persistent.
Reboot the server.
Exit Maintenance Mode on the ESXi host.
Check the PCI bus and verify that you see the VFs (with the same number of VFs on each port).
ESXi Console
# lspci -d | grep Mellanox 0000:37:00.0 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6] [vmnic4] 0000:37:00.1 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6] [vmnic5] 0000:37:00.2 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.0_VF_0] 0000:37:00.3 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.0_VF_1] 0000:37:00.4 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.0_VF_2] 0000:37:00.5 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.0_VF_3] 0000:37:00.6 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.0_VF_4] 0000:37:00.7 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.0_VF_5] 0000:37:01.0 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.0_VF_6] 0000:37:01.1 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.0_VF_7] 0000:37:01.2 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.0_VF_8] 0000:37:01.3 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.0_VF_9] 0000:37:01.4 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.0_VF_10] 0000:37:01.5 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.0_VF_11] 0000:37:01.6 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.0_VF_12] 0000:37:01.7 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.0_VF_13] 0000:37:02.0 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.0_VF_14] 0000:37:02.1 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.0_VF_15] 0000:37:02.2 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.1_VF_0] 0000:37:02.3 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.1_VF_1] 0000:37:02.4 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.1_VF_2] 0000:37:02.5 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.1_VF_3] 0000:37:02.6 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.1_VF_4] 0000:37:02.7 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.1_VF_5] 0000:37:03.0 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.1_VF_6] 0000:37:03.1 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.1_VF_7] 0000:37:03.2 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.1_VF_8] 0000:37:03.3 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.1_VF_9] 0000:37:03.4 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.1_VF_10] 0000:37:03.5 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.1_VF_11] 0000:37:03.6 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.1_VF_12] 0000:37:03.7 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.1_VF_13] 0000:37:04.0 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.1_VF_14] 0000:37:04.1 Ethernet controller: Mellanox Technologies MT28908 Family [ConnectX-6 Virtual Function] [PF_0.55.1_VF_15]
At this point you can see 16 VFs and one Physical Function (PF).
Add Network Adapter to the VM in SR-IOV Mode
Note 1: Make sure the VM version is Rel. 10 or above, and upgrade it if needed by accessing the Compatibility section (otherwise SR-IOV will not appear as an option in the network adapter selection).
Note 2: Before you start, power off the VM.
After you enable the Virtual Functions on the host, each of them becomes available as a PCI device.
To assign Virtual Function to a Virtual Machine in the vSphere Web Client:
Locate the Virtual Machine in the vSphere Web Client.
Select a data center, folder, cluster, resource pool, or host and click the Related Objects tab.
Click Virtual Machines and select the virtual machine from the list.
Power off the Virtual Machine.
Select the VM and Go to "Edit Settings".
Click on Add Network adapter.
Under Adapter Type select the SR-IOV passthrough connectivity option.
Check the Reserve all guest memory (All locked) checkbox.
I/O memory management unit (IOMMU) must reach all Virtual Machine memory so that the passthrough device can access the memory by using direct memory access (DMA).
Expand the New Network section and connect the Virtual Machine to the SRIOV net port group from the combo box at the bottom of the screen.
The virtual NIC does not use this port group for data traffic. The port group is used to extract the networking properties, for example VLAN tagging, to apply on the data traffic.
WarningMAC Address and MTU Considerations
Note 1: You can leave the automatic generated MAC address (this is the default), or change it manually.
Note 2: The Hypervisor MTU should be higher or equal to the Guest VM, otherwise, the packets may be dropped. You may modify “Set Guest OS MTU change” to allow changing MTU from guest. This step is applicable only if this feature is supported by the driver.
Power on the VM
Open the VM command line and make sure that you have the interface connected.
On the guest VM install the OS NVIDIA driver (OFED, WinOF ...).
Configure the IP Address and check Network connectivity.
Troubleshooting
At least one more VF must be configured on the firmware than on the driver. In our example we had eight VFs configured on the firmware while four are configured on the driver.
mlxconfig must be performed for each PCI device (adapter). In parallel, in the driver the configuration is per module, which means that it will be applicable for all adapters installed on the server.
Make sure the VM version is Rel. 10 or above, and upgrade it if needed by accessing the Compatibility section (otherwise SR-IOV will not appear as option in network adapter selection).
Done !
Authors
|
|
Boris Kovalev Boris Kovalev has worked for the past several years as a Solutions Architect, focusing on NVIDIA Networking/Mellanox technology, and is responsible for complex machine learning, Big Data and advanced VMware-based cloud research and design. Boris previously spent more than 20 years as a senior consultant and solutions architect at multiple companies, most recently at VMware. He has written multiple reference designs covering VMware, machine learning, Kubernetes, and container solutions which are available at the Mellanox Documents website. |