Missing network card after upgrade to PBS 3.1.4

[AraToken]

Hello!

we upgraded two of our PBS servers from 2.x to the current version 3.1.4 and while doing that we encountered an issue. Both servers are identical except that server 1 uses an additional 10G network card while server 2 got a 40G Intel card (XL710-BM2).
The first server upgraded without problems but as soon as we upgraded the second one the network card did no longer show up after performing the required reboot.

The interfaces did not show up when listing them with "ip a" or similar commands. However, lspci still listed the network card correctly. While browsing through the syslog i found the following entries:

2024-03-11T15:13:45.382845+01:00 server2 kernel: [    2.422972] i40e: Intel(R) Ethernet Connection XL710 Network Driver
2024-03-11T15:13:45.382845+01:00 server2 kernel: [    2.422980] i40e: Copyright (c) 2013 - 2019 Intel Corporation.
2024-03-11T15:13:45.382848+01:00 server2 kernel: [    2.426140] i40e 0000:04:00.0: BAR 0: can't reserve [mem 0x80800000-0x80ffffff 64bit pref]
2024-03-11T15:13:45.382849+01:00 server2 kernel: [    2.426151] i40e 0000:04:00.0: pci_request_selected_regions failed -16
2024-03-11T15:13:45.382850+01:00 server2 kernel: [    2.426213] i40e: probe of 0000:04:00.0 failed with error -16
2024-03-11T15:13:45.382850+01:00 server2 kernel: [    2.426302] i40e 0000:04:00.1: BAR 0: can't reserve [mem 0x81000000-0x817fffff 64bit pref]
2024-03-11T15:13:45.382851+01:00 server2 kernel: [    2.426315] i40e 0000:04:00.1: pci_request_selected_regions failed -16
2024-03-11T15:13:45.382852+01:00 server2 kernel: [    2.426372] i40e: probe of 0000:04:00.1 failed with error -16
2024-03-11T15:41:23.129762+01:00 server2 kernel: [    2.359296] i40e: Intel(R) Ethernet Connection XL710 Network Driver
2024-03-11T15:41:23.129762+01:00 server2 kernel: [    2.359306] i40e: Copyright (c) 2013 - 2019 Intel Corporation.
2024-03-11T15:41:23.129763+01:00 server2 kernel: [    2.359490] i40e 0000:04:00.0: BAR 0: can't reserve [mem 0x80800000-0x80ffffff 64bit pref]
2024-03-11T15:41:23.129763+01:00 server2 kernel: [    2.359503] i40e 0000:04:00.0: pci_request_selected_regions failed -16
2024-03-11T15:41:23.129768+01:00 server2 kernel: [    2.359582] i40e: probe of 0000:04:00.0 failed with error -16
2024-03-11T15:41:23.129769+01:00 server2 kernel: [    2.359670] i40e 0000:04:00.1: BAR 0: can't reserve [mem 0x81000000-0x817fffff 64bit pref]
2024-03-11T15:41:23.129769+01:00 server2 kernel: [    2.359682] i40e 0000:04:00.1: pci_request_selected_regions failed -16
2024-03-11T15:41:23.129770+01:00 server2 kernel: [    2.359743] i40e: probe of 0000:04:00.1 failed with error -16

I tried to search for similar issues in this forum but only found threads related to GPU passthrough. Booting an older kernel fixed the issue and the network card is useable again.

My question is now: are there any known issues with pci or network cards? The one we upgraded to was Linux 6.5.11-8-pve, the one currently working is the older Linux 5.15.143-1-pve. This issue also only appeared on the second server, the first one using the dual 10G card (Intel X550-T2) had no issues at all.

Thank you in advance for any help and information!

Best regards

 

[VictorSTS]

I've seen similar issues related to SR-IOV. As you are probably not using SR-IOV in a PBS server, try to disable it in BIOS. Also, check if you can upgrade nic's firmware, as it may be needed to work with a newer kernel.