Latest Proxmox update (8.2.2) renamed network interfaces

[wps_edv]

After running the update this afternoon I was shocked to find out that the server wasn't coming back. On the console I noticed that the Proxmox server is indeed running, but the network interfaces were no longer up. It looked like some of their names changed (e.g. ens259f0 into ens259f0np0).

After some investigations I found an older thread where such an "np0" attachment already happened in the past, and that I have to change /etc/network/interfaces manually to adapt to the new naming scheme. Ok, that worked...

But I wonder if Promox cannot avoid such a situation? We did a subscription to make sure that updates do not break our server (at least not in such a fatal way). Didn't anybody notice the interface renaming during the community testing phase? We have 8 interfaces. Two 10G RJ45, two 10G SFP+ and four 1G RJ45. The 10G interfaces have been renamed.

Being a Unix developer myself since decades (although more familiar with BSD than Linux) I was relatively quick to isolate the problem on the console. But I would bet such an error could ruin the weekend for some people.

 

[Darkk]

This is a known issue with major kernel updates. It's not a ProxMox thing but a Debian thing. I agree the warning of this breaking change should be pinned to the top on the what's new in PVE 8.2. There are instructions on how to pin the interface names based on mac address so it doesn't happen in the future.

Lucky it's an easy fix provided you have access to the server via other means.

 

[InfinitiGuu]

In my case, even after pinning the interface names based on MAC address, booting the new kernel still doesn't bring up the NIC interfaces.

 

[InfinitiGuu]

Looks like my issue was with the firmware level of the Broadcom 57416 10Gbase-T card. Was previously running firmware version 222.1.73.0 and when I upgraded to the latest 227.1.115.0 the networking started working. Strange that it only broke with the new 6.8 kernel.

 

[Darkk]

Yep, that would do it. One of the reasons why I upgrade the firmware on everything via the Dell's LifeCycle tool whenever I can. I will do this again before I upgrade to PVE 8.2.x as I am currently on PVE 7.4-17 on production servers at work. I am using PVE 8.2.2 at home.

 

[Ryohka233]

I was doing researches before upgrading and...though I have enp5s0 and enp4s0, IP address is actually assigned to vmbr0(which has a slave enp4s0) on my setup? is that normal?

 

[_gabriel]

vmbr0(which has a slave enp4s0) on my setup? is that normal?

yes, PVE works out the box with a bridge vmbr0, because VMs need bridge to have its network bridged to vmbr0.

 

[Ryohka233]

yes, PVE works out the box with a bridge vmbr0, because VMs need bridge to have its network bridged to vmbr0.

Thank you for the explanation. So I assume if my network interface name changed, I just make whatever changes I need in /etc/network/interfaces? As vmbr0's bridge port is enp4s0 now.

 

[supermicro_server]

In my case, I tried to install PVE8.2 in a new server but the AIOM network card are not recognised by the system.

If I try to upgrade the system from PVE8.1 to PVE8.2 the network card doesn't work at 100%: I have to restart the networking service manually.
Very tragic situation...

 

[kyesil]

Hi ;
Why auto replace interface names during upgrade. I am shoked. Connection lost remote dedicated server. There is no warning about this. It does not make sense.

 

[helptec]

For me happened the same after upgrade from 8.1 to 8.2
Two interfaces has been renamed

enp129s0f0  ->   enp129s0f0np0
enp129s0f1  ->   enp129s0f1np1

 

[rkb405]

Same thing here, luckily it was not on a production system, there has to be a way around this update issue