Connext-6 lx adapter dies

Infrastructure & Networking Adapters and Cables Ethernet Adapter Cards
1

Hi,
I just get a double ports Connext6-lx board (MCX631102AS-ADAT).

I installed it on a Windows10 PC among with other fully working Mellanox adapters (Connext4-lx and Connext6-dx) to compare the performance.

I downloaded the driver yesterday.

Anyway
The boards seems to be installed correctly, but as soon as I change any driver settings for one the port (for instance I need jumbo frame) the port dies and I need to reboot the PC to get it working agains. This happens for any settings or once I try to use the board to grab data from cameras (I can grab data using any other ports of the other two Mellanox boards).

These are the infomation from the dead port

image
image374Ă—433 56.8 KB

These are the other one that I still do not touch (sorry I can attach just one screenshot)

Driver Version : 25.1.26647.0
Firmware Version : 26.39.3004
Port Number : 2
Bus Type : PCI-E 16.0 GT/s x8
Link Speed : 10.0 Gbps/Full Duplex
Part Number : MCX631102AS-ADAT
Serial Number : MT2447J00NEK
Device Id : 4127
Revision Id : 0
Current MAC Address : B8-E9-24-C8-95-83
Permanent MAC Address : B8-E9-24-C8-95-83
Network Status : Connected
Adapter Friendly Name : Ethernet 6
Port Type : ETH
IPv4 Address #1 : 169.254.28.84
IPv6 Address #1 : fe80::8c93:8564:3a4d:936%18

Once I changed the settings I got in the general tab of the device settings in Device Manager

This device cannot start. (Code 10)

This device does not exist.

Note if I scan for hardware changes in the Device Manager both ports disappear.

Usually after reboot it seems to work, but the last time, windows stops it with this error
Windows has stopped this device because it has reported problems. (Code 43)

Event viewer reports these messages

  • AD-1: FW health report - fw ver 26.39.3004, hw 0x216, rfr 0x0, callra 0x208c2500, var[0] 0x73, var[1] 0x73, var[2] 0x0, var[3] 0x0, var[4] 0x0, synd 16, ext_synd 0x0, exit_ptr 0x208c1db0
  • AD-2: FW health report - fw ver 26.39.3004, hw 0x216, rfr 0x0, callra 0x208c2500, var[0] 0x73, var[1] 0x73, var[2] 0x0, var[3] 0x0, var[4] 0x0, synd 16, ext_synd 0x0, exit_ptr 0x208c1db0
  • The network interface “Mellanox ConnectX-6 Lx Adapter” has begun resetting. There will be a momentary disruption in network connectivity while the hardware resets. Reason: The network driver detected that its hardware has stopped responding to commands. This network interface has reset 1 time(s) since it was last initialized.
    -ConnectX-6 Lx device reports a send completion handling timeout on TxQueue 0xFFFFC30CAD62C000, Cq is armed. Attempting recovery.
    -NDIS initiates reset on device ConnectX-6 Lx.
    -ConnectX-6 Lx: OID Statistics:
    Last OID: 0x1010103, took: 5 micro second.
    Most time consuming OID 0x20106, took: 1011611 micro second.
    -ConnectX-6 Lx: dump was created at folder (dmn-GN-TXCQ-NA-2025.2.13-11.9.39-Gen0) due to dump-me-now request with source TX CQ STUCK.
    Dump-me-now dumps are placed by default in folder %SystemRoot%\temp\Mlx5_Dump_Me_Now-193-0-0
    or folder that was set by registry keyword HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Class{4d36e972-e325-11ce-bfc1-08002be10318}<nn>\DumpMeNowDirectory

Is the board faulty or I miss some obvius point?

If you need I can upload the dump
G.

2

The problem seems to be related to the temperature.

I compare the Connext6-dx and the Connext6-lx that I have.

The DX has a heatsync three time larger than the LX and then the latter easily reach the temperature limit (around 120°C) and the firmware itsself switch off the board.

I made some test with a “powerful” fan in the chassis.

To keep the board under 100°C with 25°C of ambient, I need a fan 2 or 3 times faster than with other adapters (either Intel or older Mellanox) i.e. from 1500rpm to 3000rpm or more.

IMHO the heat dissipation in this board was not designed correctly!!!

1 Like
3

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.