Problem e1000 driver

[Dexter23]

Hi all

Proxmox: 7.1-8

Kernel: 5.13.19-2-pve

From last few months i have problem with network adapter, the server has two ethernet adapter one is integrated on motherboard and the other is pci-e cards, the problem is only on integrated ethernet adapter.

And the error on the LOG is this:

Dec 23 10:04:41 pve kernel: e1000e 0000:00:1f.6 eno1: Detected Hardware Unit Hang:
TDH <ce>
TDT <c>
next_to_use <c>
next_to_clean <cd>
buffer_info[next_to_clean]:
time_stamp <103a911c7>
next_to_watch <ce>
jiffies <103a91a38>
next_to_watch.status <0>
MAC Status <80083>
PHY Status <796d>
PHY 1000BASE-T Status <3800>
PHY Extended Status <3000>
PCI Status <10>
Dec 23 10:04:42 pve kernel: e1000e 0000:00:1f.6 eno1: Reset adapter unexpectedly


How can i fix that?

Thanks you

 

[oguz]

hi,

it's a known issue with intel NICs, see here [0]

workaround: thtool -K eno1 gso off gro off tso off tx off rx off (eno1 is the interface name)


[0]: https://forum.proxmox.com/threads/e1000-driver-hang.58284/#post-279338

 

[Dexter23]

Hi

but is it possible that a patch has not yet been made to fix this problem?

I have to propose to my customers to switch from Vmware Esxi to proxmox but at this point since their servers will surely have intel as network cards I don't know if I should migrate them from Esxi to Proxmox

By doing this workaround I have seen that performance on data transfer speed is lost

 

[oguz]

but is it possible that a patch has not yet been made to fix this problem?

you can follow the upstream bug report here [0], but currently not fixed..
there's no single patch to fix the problem, it was fixed but came up again in different versions or different hardware, modern firmware and hardware is complex

I have to propose to my customers to switch from Vmware Esxi to proxmox but at this point since their servers will surely have intel as network cards I don't know if I should migrate them from Esxi to Proxmox

there are other NICs as well, they're are not that expensive...

By doing this workaround I have seen that performance on data transfer speed is lost

where have you seen that bit about the performance loss?

quite a few people in the forum tried the workaround and it didn't seem to cause any obvious bottlenecks with disabled offloading so far (at least no reports that i've seen)... but if you can get some numbers would be interesting to check

[0]: https://bugzilla.kernel.org/show_bug.cgi?id=205047

 

[Ouji]

Disabling offloading doesn't resolve for me. I had an install with Proxmox 6.1 that I had to purge and this was a non-issue, so this is very weird. Might need to ditch Proxmox as a system since purchasing a new NIC is not an option for me. It's a real shame, because I really like Proxmox

 

[Dexter23]

Hi all

I don't waste my time to do this workaround so i install new ethernet pci card on the server.

Thanks anyway for the support