error i40e NIC Link is Down

[gly]

Hello,

For a week now we have more and more the following error:

Mar 29 08:09:01 pve systemd[1]: Started Proxmox VE replication runner. Mar 29 08:09:17 pve kernel: i40e 0000:64:00.0 eno1: NIC Link is Down Mar 29 08:09:17 pve kernel: vmbr0: port 1(eno1) entered disabled state Mar 29 08:10:00 pve systemd[1]: Starting Proxmox VE replication runner... Mar 29 08:10:01 pve systemd[1]: pvesr.service: Succeeded. Mar 29 08:10:01 pve systemd[1]: Started Proxmox VE replication runner. Mar 29 08:10:57 pve kernel: i40e 0000:64:00.0 eno1: NIC Link is Up, 1000 Mbps Full Duplex, Flow Control: None Mar 29 08:10:57 pve kernel: vmbr0: port 1(eno1) entered blocking state Mar 29 08:10:57 pve kernel: vmbr0: port 1(eno1) entered forwarding state Mar 29 08:10:57 pve kernel: i40e 0000:64:00.0 eno1: NIC Link is Down


When this happens, we can no longer access Proxmox and VMs.

I searched on the forum and found that the driver version of the card may be involved, on kernel 5.4.73 the version is:

filename: /lib/modules/5.4.73-1-pve/kernel/drivers/net/ethernet/intel/i40e/i40e.ko version: 2.8.20-k

The version available from Intel is 2.14.13. I have read that if we update the driver it might roll back to the old version with the PVE update.

What is the right solution please?

Thanks a lot

 

[Stoiko Ivanov]

i40e 0000:64:00.0 eno1: NIC Link is Down

This would look like the NIC loses the Link - possible causes:
* problem in the driver - please try to install the latest pve-kernel first and check if this fixes the issue (I personally would not recommend to install the out-of-band intel drivers
* check the complete journal/dmesg for messages from i40e and others which might give a hint at what's not working
* actual problem of the wiring - change the cable/fibre between the NIC and the switch
* some problem on the switch - check the logs of the switch (if available) - else try a different port or different switch

I hope this helps

 

[gly]

This would look like the NIC loses the Link - possible causes:
* problem in the driver - please try to install the latest pve-kernel first and check if this fixes the issue (I personally would not recommend to install the out-of-band intel drivers
* check the complete journal/dmesg for messages from i40e and others which might give a hint at what's not working
* actual problem of the wiring - change the cable/fibre between the NIC and the switch
* some problem on the switch - check the logs of the switch (if available) - else try a different port or different switch

I hope this helps

We had already tested to update the kernel and we tested the pve-test version. In view of the urgency and the fact that the server supports production VMs, we migrated from Proxmox to ESXi. The ticket may be closed. Thks

 

[filipealvarez]

I'm having exactly this issue with i40e / Intel 10Gb card on latest Proxmox 6.4 / Kernel 5.4.

Somenone could help with sugestions ? It was working for 2 years with 5.0 Kernel but after upgrade this "Link Down" starts...

 

[xed]

Posted my own here:
https://forum.proxmox.com/threads/n...-i40e-nics-integrated-with-dac-cables.152939/

 

[xed]

Do you have ASPM enabled for PCIe?