Mellanox ConnectX-2 not visible in network devices

[rb28dett]

Hello folks,

I'm running unRaid 6.6.3 and I have a Mellanox ConnectX-2 installed in my system, and the webui's system devices shows it as recognized:

 

[15b3:6750] 09:00.0 Ethernet controller: Mellanox Technologies MT26448 [ConnectX EN 10GigE, PCIe 2.0 5GT/s] (rev b0)

 

Clearly, the system sees it and has the right driver for it. However, the device doesn't show up in the network options menu so I can't actually set it up and assign an IP to it.

 

The only thing I can think of is that I have set up a 4-port LACP aggregate using the regular 1Gb ports on the server, and maybe having a bond enabled somehow doesn't allow a new NIC to show up.

 

I haven't tried breaking up that bond yet into separate individual ports again to see if the Mellanox card will show up since its a pretty messy operation.

 

Any idea on what to do next to get the Mellanox to show up in the network settings? Is there something I'm missing here?

[Squid]

35 minutes ago, rb28dett said:

Clearly, the system sees it and has the right driver for it.

Doesn't necessarily mean a driver is loaded for it.  

 

What does

lspci -k

 

show?

[rb28dett]

Sorry I should've mentioned I checked that it was loaded with the mlx4_core driver.

 

Relevant output of 'lspci -k'

 

09:00.0 Ethernet controller: Mellanox Technologies MT26448 [ConnectX EN 10GigE, PCIe 2.0 5GT/s] (rev b0)
        Subsystem: Mellanox Technologies MT26448 [ConnectX EN 10GigE, PCIe 2.0 5GT/s]
        Kernel modules: mlx4_core

 

Edited November 2, 2018 by rb28dett
clarity

[JorgeB]

7 hours ago, rb28dett said:

The only thing I can think of is that I have set up a 4-port LACP aggregate using the regular 1Gb ports on the server, and maybe having a bond enabled somehow doesn't allow a new NIC to show up.

This shouldn't be a problem, please post your diagnostics, maybe something visible there: Tools -> Diagnostics

[rb28dett]

Diagnostics attached

tower-diagnostics-20181102-1607.zip

[JorgeB]

NIC is found but driver isn't loading:

 

Nov  1 17:41:50 Tower kernel: mlx4_core 0000:09:00.0: command 0xff6 failed: fw status = 0xb
Nov  1 17:41:50 Tower kernel: mlx4_core 0000:09:00.0: RUN_FW command failed, aborting
Nov  1 17:41:50 Tower kernel: mlx4_core 0000:09:00.0: Failed to start FW, aborting
Nov  1 17:41:50 Tower kernel: mlx4_core 0000:09:00.0: Failed to init fw, aborting.
...
Nov  1 17:41:50 Tower kernel: mlx4_core: probe of 0000:09:00.0 failed with error -14

Bad NIC? Or possibly some incompatibility with the board/slot, you can try googling the error.

[rb28dett]

A little late but I figured I'd let anyone know if they ran into the same issue. One of the NICs was bad - couldn't re-flash the firmware either. I got it replaced by the seller and all is well now.

[mrbilky]

On 11/1/2018 at 11:04 PM, rb28dett said:

Sorry I should've mentioned I checked that it was loaded with the mlx4_core driver.

 

Relevant output of 'lspci -k'

 

09:00.0 Ethernet controller: Mellanox Technologies MT26448 [ConnectX EN 10GigE, PCIe 2.0 5GT/s] (rev b0)
        Subsystem: Mellanox Technologies MT26448 [ConnectX EN 10GigE, PCIe 2.0 5GT/s]
        Kernel modules: mlx4_core

 

Just curious I don't see the driver in your screen shot as it shows in mine? Is that normal

 

01:00.0 Ethernet controller: Mellanox Technologies MT26448 [ConnectX EN 10GigE, PCIe 2.0 5GT/s] (rev b0)
        Subsystem: Mellanox Technologies HP 10 GbE PCI-e G2 Dual-Port NIC (rev C1)
        Kernel driver in use: mlx4_core
        Kernel modules: mlx4_core

[rb28dett]

On 11/27/2018 at 4:13 PM, mrbilky said:

Just curious I don't see the driver in your screen shot as it shows in mine? Is that normal

 

01:00.0 Ethernet controller: Mellanox Technologies MT26448 [ConnectX EN 10GigE, PCIe 2.0 5GT/s] (rev b0)
        Subsystem: Mellanox Technologies HP 10 GbE PCI-e G2 Dual-Port NIC (rev C1)
        Kernel driver in use: mlx4_core
        Kernel modules: mlx4_core

No its not normal. My output says it has the kernel driver 'mlx4_core` but it doesn't say it has loaded it. That's where the issue was. Yours is expected and normal output.

[Alanjm]

I am having the same issues, the card recognised but mlx4_core is showing Kernel driver in use. any ideas how I can resolve this?

The card is not showing in the networks tab.

 

Ethernet controller: Mellanox Technologies MT26448 [ConnectX EN 10GigE, PCIe 2.0 5GT/s] (rev b0)
        Subsystem: Mellanox Technologies MT26448 [ConnectX EN 10GigE, PCIe 2.0 5GT/s]
        Kernel driver in use: mlx4_core
        Kernel modules: mlx4_core

Edited November 4, 2019 by Alanjm