ConnectX®-4 Single 100Gb MCX455A-ECAT has Ref. Def. F3 fuse blown

Hi

We have an issue with ConnectX®-4 Single 100Gb MCX455A-ECAT cards, that has 12V fuse (F3) blown on 3 cards

The set up is below:

  1. Mainboard with two x16 PCIE Gen3 slots

  2. Two MCX455A-ECAT cards installed on these two slots

  3. Use a copper cable MCP-1600 to connect these two card together

  4. OS: RHEL/Centos 7.3 aarch64

  5. Test system put into a LAB with room temp around 20 degree C, Force air provided for cards with the direction from Mellanox Chip set to port, airflow speed >450 LFM.

Running iperf test or ping test for long time, and see that there were 3 cards failed with the same phenomenon that the 12V fuse F3 being burnt.

We don’t know what is the possible cause of the issue, whether:

  • The fuse rating is not enough to handle the Max current in some circumstance?

  • The specific card has power consumption higher than expected?

  • Or any other cause that we have not aware of??

Appreciated if one has encountered with similar issue & share the information

Thanks

Hi Triet,

Base on what you described, there are a lot of factors and open questions on hand. It’s best you open a support ticket with Mellanox to investigate further.

Please send an email to the below email and include step by step details to reproduce.

support@mellanox.com mailto:support@mellanox.com

Cheers,

~Rage

Hi all,

Is there any comment?