network card not visible, driver or hardware error?

P

proxtib

Well-Known Member
Dec 16, 2020
30
1
48
35
Hello,

I use the Asrockrack X570D4I-2T (driver are somthing like X550 or ixgbe ? )
I had to take the card out of the case, for a little cleaning (dust) check for disk problems.
Once this is done, Proxmox (pve) starts but no trace of the 2 integrated network cards.
No problem with the IPMI card

With root user:

Code: 
#   pveversion
pve-manager/9.1.4/5ac30304265fbd8e  ( running kernel: 6.17.4-2-pve)
#  lspci | grep -iE  'eth|net|intel'
#  ls /sys/class/net/
bonding_masters enxmac@1 lo
# lsmod | grep -Ei 'X550|ixgbe'
# dmesg | grep -iE 'enx7|ixgbex550|err'   # I resume logs
cdc_ether 3-5.3:2.0  enxmac@1: renamed from usb0 # USB0 ???
GPT: Use GNU Parted to correct GPT errors.

with ip link, I see my bonds, lo and this enxmac@1 (new interface ?) but not my previous enp35s0f0 and enp35s0f1.

I have 2 zfs pool, both are degraded -> raidz1, both have one 1 one disk or ssd ko ( or the probleme is with the OCuLink cables ? )
For disks, I will try to do tests with other disks to see if it is a cable problem or really a disk problem.

I would already like to know if in your opinion it is a hardware problem with the integrated network cards or something else.

Thanks
 
Last edited:
If the system was working previously and the NICs were being detected, then it is more likely than not that you are dealing with a hardware or firmware issue.

You may need to try a different PCIe slot for the cards you removed, or temporarily remove all non-essential expansion cards to establish a clean baseline. It is also possible that something was reset or failed during the process.

As a next step, you could create a CD/USB with installation media, disconnect all non-essential devices, and check whether booting from the ISO detects the integrated NICs. Also note that the new interface is USB-connected, was this present in the previous configuration?

You can also try to reset the BIOS.

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
 
Check in the BIOS setup that your hardware is detected there maybe? Usually you can setup PXE or some options (like.. disabling them) in the setup..
If you don't see anything in dmesg, linux probably doesn't see it at all anyway. You can still try to check dmidecode or lshw, but I'd say there probably won't be anything there either..
 
Damn, you confirm my fears...

Both cards are soldered to the motherboard PCB. So I can't take them off like that.

No it wasn't there, but it may be the USB network card that I connected. But when executing the 'ls /sys/class/net/' command, I restarted proxmox and my usb card was not connected; so why does it appear with this command?

Regarding the tests with ISO, I already did it yesterday. It seems to me that I also see this xxx usb card (even though it is not connected) but see nothing else.

I will test the bios reset and the test with the iso.

But when I go to the bios, I have the possibility of activating these network cards or not. If they are ko, that is not possible, right? I should have an error message, or a flashing LED, right?


----

with lshw -class network, I see

Code: 
*-network DISABLED
    description : Ethernet interface
    physical id: 4
    bus info usb@3:5.3
    logical name enxmac@1
    serial   .......
    capabilities: ethernet physical
 
Last edited:
So with a iso (archlinux iso) I see

Code: 
# ls /sys/class/net
enp42s0f3u5u3c2 lo
# cat /sys/class/net/enp42s0f3u5u3c2
a MAC @; the same as  enxmac@1  proxmox card

With proxmox


Code: 
# lshw -class network
*-network DISABLED
    description : Ethernet interface
    physical id: 4
    bus info usb@3:5.3
    logical name enxmac@1
    serial   .......
    capabilities: ethernet physical

I do not see any network information with 'dmidecode' (both proxmox and arch iso)

---
Question may be stupid but I have a doubt:
If I take another X570D4I-2T ; that I reconnect any disk, cpu, ram; the operating system see the difference, will make mistakes?
 
Last edited:
First, none of the challenges you are describing are related to the Proxmox Virtual Environment.

To clarify: dusting your PC was not a contributing factor, and the on-board integrated NICs never worked previously.

It is possible that the NICs are dead or faulty, but that is unlikely. A more probable explanation is that they use an uncommon chipset that is not supported by the PVE kernel or by Linux in general.

If I were in your position, I would:
a) Remove the USB network adapter and set it aside, as it appears to be introducing additional confusion.
b) Boot a vanilla Debian or Ubuntu installation ISO (or a rescue ISO) and verify whether the NICs are detected by those kernels. Make sure the NICs are explicitly enabled in the BIOS.

If the NICs are detected under vanilla Debian, proceed by installing PVE as packages on top of the Debian OS.



Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
 
The NICs worked very well until the beginning of the week.

I unplugged all the disks and plugged in a new one.

When I boot from the USB key with Debian 13.3.0 ISO, I can see an enx@mac1 network interface but it does not seem to work.

Debian was therefore installed without networks.

First, with some research:

- Interfaces enXXXXXXXX are based on the mac address of the card. XXXXXXXXXX is a mac address.
- Interfaces enpXXXXXX are basedd on the PCI bus. enp35s0f0 say: en:Ethernet; p35: PCI bus number 35, s0: slot 0, f0: fonction 0
And enp35s0f0,the same but fonction 1

I have replace the real Mac address with AddressMAC1.

From this new installation, I found this enxAddressMAC1 interface. AddressMAC1 is therefore its mac address, the I configured the networks by hand, but the interface still doesn't work.

I don't understand, the built-in interfaces worked very well.
Seen from proxmox they were named enp35s0f0 and enp35s0f1, their names were therefore not based on their MAC address but on their PCI bus.

So PCI buses are no longer visible from the OS and why now I have this enx interface ?

But there is still this enx@mac1 interface.
And it's not the Mac address of my USB network card, I checked.
On the other hand, my USB network card works.

I contacted Asrock support. He told me to update the motherboard fimwarre. It's done, but it doesn't change anything.

Doesn't that look like a driver problem?
 
Last edited:
proxtib said:
Doesn't that look like a driver problem?
Click to expand...
Network cards are driven by chipsets - the “brains” of the NIC. Most of these chipsets are designed and manufactured by a small number of companies and then licensed or sold to vendors who do not fabricate their own silicon. Mellanox and NVIDIA are examples of companies that do design and manufacture their own chipsets.

Companies like ASRock do not have fabrication facilities to manufacture chipsets for NICs, GPUs, or even motherboards. Instead, they source chipsets from third-party manufacturers and assemble finished products around them. This is still a high-tech process, but it is very different from actually designing and fabricating silicon.

Because most NIC chipsets come from a small set of vendors (Intel, Realtek, Broadcom, etc.), they have well-defined and standardized behavior. As a result, the Linux community typically only needs to implement a single driver per chipset family, which then supports hundreds of NIC models across many vendors.

These drivers are generally included directly in the Linux kernel. In rare cases, highly customized or proprietary chipsets may require vendor-supplied drivers. Sometimes those drivers must be installed as packages or even compiled from source. It is unlikely that this applies to your situation.

Can this be a driver problem? In theory, yes, anything is possible. Is it likely? No.

What is more likely is a system integration issue. I.e. the combination of components in your system (motherboard, disks, NICs, GPU, etc.) is not playing well together. If your disks are NVMe, they are PCIe devices. Your onboard NICs are PCIe devices. Everything is sharing the same PCIe bus provided by the motherboard. Conflicts or incompatibilities can and do occur, especially on non-enterprise hardware, where compatibility can be something of a coin toss.

As suggested earlier, remove as many components as possible and simplify the system. Boot from a vanilla Ubuntu or Debian ISO/USB and observe which devices appear or disappear. Keep careful notes. You may also want to try an Ubuntu HWE kernel.

You could go further by compiling a custom kernel and experimenting with PCI-related options and drivers, but that is well beyond the scope of this forum.

PS its also possible, if those disks you pulled are NVMe, that you damaged the PCI connectivity.

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
 
You must log in or register to reply here.
Top Bottom
Back