Pf, sf and their representors disappeared after installation DOCA 2.2.0/DOCA 2.5.0 both in DPU and host

Infrastructure & Networking BlueField
1

hi everyobe , i am in trouble, that make crazy! the problem is： I have 2 same dpus which type is MBF2H332A-AENO_Ax_Bx, they are installed in the same sever . We have some app developed base on DOCA 2.2.0 , so i update DOCA to 2.2.0, but
1、on host side, wo can find any interface from dpu, and kernel message have errors like this:

image
image1357×384 15.4 KB

2. on arm/dpu side, ifconfig show like this:
image
image1034×720 26.1 KB

and dmesg show like this :
image
image1300×240 16.3 KB

the representors to host pf0hpf adn pf1hpf disappeared, the representors to arm and sf disappeared too! ovs can not start too!
I installing full DOCA image on dpu via host , and repeated that many times, but the problem occurs always!
howerver, old DOCA versions work well ,such as, DOCA 2.0.2 , DOCA 1.5.x, DOCA 1.3.x.
dose any one have problem like this ? i am crazy

2

Hi,
I use Doca 2.2.0 and there is no such problem

3

hi, bro, in my laboratory, l have an sever which is some type of DELL, and only on dpu was installed in it, DOCA 2.2.0 work well, i programmed and tested my apps smoothly! But the in data centre, we have four servers which is not dell, and 2 dpus are installed on each each server , then , i got this probelm!
Dose each server can take only one dpu?

4

I think no, I used 2 DPU in one server

5

can you share firmware version?

6

DPU’s firmware version？the bfb is DOCA_2.2.0_BSP_4.2.0_Ubuntu_22.04-2.23-07.prod.bfb
and DOCA 2.2.0 built-in firmware is 24.38.1002

7

and, when i try to update this firmware , i got this error:

image
image1778×801 27.5 KB

8

You use 24.32.2004 version, you should perform power off cycle (power off server and turn it after that, not reboot)

9

You can take mlxofedinstall tool with force-fw-update option and perform power cycle after that

10

version 24.32.2004 was installed with DOCA 1.3.x , ww install that version a year ago , now, we need update to DOCA 2.2.0 and its firmware is 24.38.1002！ I use IBMC control the server remotly, I click the button “Force power off and then power on again” to restart the server

11

can you share update.log file? Do you see the error reason?

12

is it xFusion server?

13

the log is same as screenshot , it says
image

14

YES, it is

15

so, have you encountered similar problems before?

16

try to upgrade the DOCA BSP to version 1.5 first and if everything will be ok then on version 2.0.2

17

Yes, I have

18

I have tried , DOCA 1.5.X and DOCA 2.0.2 work well, no problem! But DOCA 2.2.0, DOCA 2.5.0, the problem appears! But we need 2.2.0, our apps was develop by DOCA 2.2.0！now the apps must online , but the problems trouble us !

19

any solution could you tell, do we have to give up 2.2.0 or the more new version?

20

I think you should update firmware up to 24.38.1002 and check that this version is running.
you can use MLNX_OFED_LINUX-23.07-0.5.1.2-ubuntu22.04-x86_64 driver and mlxofedinstall tools with --force-fw-update --fw-update-only options.
Now I see that you use very old firmware.
the problem with xFusion it’s separate story, now I can’t help you. Maybe later when I will get more details.