Server Will not Boot - MCX755106AS-HEAT

I have multiple ConnectX-7 (OPN = MCX755106AS-HEAT) devices in a GEN4 PCI Express chassis. If I allocate one to a server the server fails to boot. It doesn’t matter which device is allocated, the result is the same.

I have updated the firmware on these devices to version 28.47.1088 (the previous version was 28.44.1036).

Has anyone experienced this kind of an issue? If so, how did you resolve it?

Thanks.

Hi

It looks like CX-7 H/W fault.

in case HCA H/W fault, sometimes the server fails to boot up.

Did you ever try with another new one?

with new CX-7, if the problem persists, please open a CASE.

/HyungKwang

Hi hyungkwanc,

Thanks for the reply. We have eight of these CX-7 NICs. We have tried all eight of them. They all cause the same problem. With the previous FW version (28.44.1036) and the newest FW version (28.47.1088).

Are you saying the cards are no good?

– Lynn

Hi Lynn

Did you ever try&test 8 HCAs on another server?

Based on my experience, bad HCA can cause boot-failure. But it’s not normal that all 8 HCAs interrupt boot-failure.

Did you install OFED or DOCA in a server? other than built-in?

I’d like you to test it on another server. i think it’s not a FW related issue.

/HyungKwang

Hi HyungKwang,

Thanks again for the reply. We have tried this in three other servers with the exact same result. With each server we’ve allocated each one of the ConnectX-7 NICs and gotten the same result.

We have taken all eight ConnectX-7 NICs out of the expansion chassis and plugged each one directly into a PCI Express slot in yet another server. In this case the server boots and recognizes each ConnectX-7 NIC. This is the setup we used to update the firmware on the NICs.

We have other PCI Express devices plugged into the PCI Express expansion chassis. We have allocated these devices to the four servers one at a time and the server always boots and recognizes the device allocated to it. We have also successfully allocated more that one of these devices to each server and that also works.

Any other things we can try?

– Lynn

going back to your initial description, you put CX-7 to Gen-4 Chassis.

Basically, ConnectX-7 supports PCIe Gen 5.0, of course compatiable with PCIe Gen 2/3/4.

What about testing it with genernal Gen5.0 Server ?