No network after upgrade to version 8

[techopsguy]

Hello guys! I found myself in a tight spot.

Yesterday, my galaxy brain decided that it would be nice to upgrade Proxmox 7 to 8 remotely. I have a VPN set up in my router, and I've been remoting in with no problem for the past 2 years (server is 3000km away). I started playing around like many other times and I found out I was using an outdated version, how do I dare!
So I went to tteck and started with the preparatives, backed up the couple important VMs and all set. After that, I lost connectivity.

I have no idea why, the new system (kernel 6.2) won't boot. It gets stuck right on "Loading Linux 6.2.16-3-pve".(https://forum.proxmox.com/threads/hangs-on-screen-loading-linux6-2-13-pve.129355/ probably related)
So, with the aid of my completely non tech dad, I've managed to at least login into it by using an old kernel (5.13). From there, I said, well if I can start that, I can connect, well I was wrong because apparently the network interfaces won't go up. I've assisted my dad (he actually has assisted me!) spelling every command I can think of, and my network just won't get up.
In the beginning I thought the error was the typical PCI naming and swap of the interfaces name, so I changed that.
ip -a would show enp4s0 and enp5s0 (I have two network cards, motherboard and an extra one) and in fact, in /etc/network/interfaces I had enp5s0 and enp6s0, so I changed that accordingly. Rebooted many times, but the thing is still not up. vmbr0 doesn't show when hitting 'ip a'
I know the interfaces file by heart now, as it's really basic/standard.

auto lo
iface lo inet loopback

iface enp4s0 inet manual
iface enp5s0 inet manual
auto vmbr0
iface vmbr0 inet static
address 192.168.1.5/24
gateway 192.168.1.1
bridge-ports enp4s0
bridge-stp off
bridge-fd 0

systemctl restart networking won't do anything, neither rebooting the server, but service is active and running. Only enp4s0 and enp5s0 showing up in 'ip a' with state DOWN. And physically, there's no led/link in any of the adapters. Tried using ifreload -a and I get permission denied... (maybe because of sloppy upgrade? No clue)
I even tried to setup one of the interfaces with dhcp to see if maybe I get connectivity and I can work the bridge with the other one, but nothing.

What are my options? What am I missing?

Thanks for your help in advance!

 

[jsterr]

Hello guys! I found myself in a tight spot.

Yesterday, my galaxy brain decided that it would be nice to upgrade Proxmox 7 to 8 remotely. I have a VPN set up in my router, and I've been remoting in with no problem for the past 2 years (server is 3000km away). I started playing around like many other times and I found out I was using an outdated version, how do I dare!
So I went to tteck and started with the preparatives, backed up the couple important VMs and all set. After that, I lost connectivity.

I have no idea why, the new system (kernel 6.2) won't boot. It gets stuck right on "Loading Linux 6.2.16-3-pve".(https://forum.proxmox.com/threads/hangs-on-screen-loading-linux6-2-13-pve.129355/ probably related)
So, with the aid of my completely non tech dad, I've managed to at least login into it by using an old kernel (5.13). From there, I said, well if I can start that, I can connect, well I was wrong because apparently the network interfaces won't go up. I've assisted my dad (he actually has assisted me!) spelling every command I can think of, and my network just won't get up.
In the beginning I thought the error was the typical PCI naming and swap of the interfaces name, so I changed that.
ip -a would show enp4s0 and enp5s0 (I have two network cards, motherboard and an extra one) and in fact, in /etc/network/interfaces I had enp5s0 and enp6s0, so I changed that accordingly. Rebooted many times, but the thing is still not up. vmbr0 doesn't show when hitting 'ip a'
I know the interfaces file by heart now, as it's really basic/standard.

auto lo
iface lo inet loopback

iface enp4s0 inet manual
iface enp5s0 inet manual
auto vmbr0
iface vmbr0 inet static
address 192.168.1.5/24
gateway 192.168.1.1
bridge-ports enp4s0
bridge-stp off
bridge-fd 0

systemctl restart networking won't do anything, neither rebooting the server, but service is active and running. Only enp4s0 and enp5s0 showing up in 'ip a' with state DOWN. And physically, there's no led/link in any of the adapters. Tried using ifreload -a and I get permission denied... (maybe because of sloppy upgrade? No clue)
I even tried to setup one of the interfaces with dhcp to see if maybe I get connectivity and I can work the bridge with the other one, but nothing.

What are my options? What am I missing?

Thanks for your help in advance!

Maybe need auto enp4s0 and auto enp5s0 like you have for vmbr0. Also check if enp4s0 is the correct interface with correct mac-address you want to have on your vmbr0 (it could also be enp5s0, when there was name changes in udev).

 

[techopsguy]

Hi jsterr! Thanks for replying.

I may have input the interfaces wrongly while doing it by memory! I used the template from the wiki to post it here, I will take a picture later (if my dad allows more pestering haha) but it's possible they're on auto.

Regarding the interface change, I've tried with both (enp4s0 and enp5s0) as I have both adapters wired, but no luck.

Even with all that info, wouldn't at least the interface I set for dhcp work? It just won't get "UP".

I should point, I think the adapters are both realtek. Am I cursed with 'Realtekitis disease'? People usually get their interfaces up, and then they shutdown for some reason after a certain time. In my case they never go up. Could it be related?

Thanks again!

 

[jsterr]

Hi jsterr! Thanks for replying.

I may have input the interfaces wrongly while doing it by memory! I used the template from the wiki to post it here, I will take a picture later (if my dad allows more pestering haha) but it's possible they're on auto.

Regarding the interface change, I've tried with both (enp4s0 and enp5s0) as I have both adapters wired, but no luck.

Even with all that info, wouldn't at least the interface I set for dhcp work? It just won't get "UP".

I should point, I think the adapters are both realtek. Am I cursed with 'Realtekitis disease'? People usually get their interfaces up, and then they shutdown for some reason after a certain time. In my case they never go up. Could it be related?

Thanks again!

Could also help to post the output of "dmesg" and the output of "lshw -c network -businfo" You can see what the adapters names are by typing ip a and you can try to blink the adapter with ethtool -p enp4s0 if you have someone onsite.

I never used realtek, but we might find some info if you post the lshw output (if it is installed), You can also post lspci -vvv | grep Realtek

 

[techopsguy]

I'll try getting him around in 1h or so. Anything else I should check for? I will ask him to take pictures of everything I find useful.

Thanks a lot again!

 

[techopsguy]

Okay, first of all, I think if we keep doing this, my dad is gonna end up being a Linux connoisseur. The guy has qualities!

The upgrade doesn't allow me to boot with Linux 6.2 kernel. It gets stuck at "loading Linux 6.2.13-pve...". So I have to choose the only bootable kernel that is 5.15.108-1pve.

The /etc/network/interfaces file is exactly like this:

iface lo inet loopback

auto enp4s0
iface enp4s0 inet dhcp (this is the one I left out of the bridge to see if it could go up, but nah)

auto enp5s0
iface enp5s0 inet manual

auto vmbr0
iface vmbr0 inet static
address
gateway
netmask
bridge-ports enp5s0
bridge-stop off
bridge-id 0
dns-nameservers

Addresses and all that are correct, that part of the config has not been touched.

-------
Tried showing stuff with lshw but command not found (should this be normal?)
ethtool -p enp4s0/5s0 says: Cannot identify NIC: Operation not supported (I don't know how fudge this installation is, to be fair)

lspci shows both ethernet contollers fine with RTL8111/8168/8411, but then both load r8169 driver.
dmesg | grep enp only shows "enp5s0 renamed from eth1 and enp4s0 renamed from eth0"

I tried blacklisting 8169 driver and updating initramfs, so maybe it would revert to 8168, but then they won't even show up in "ip a".

Due to not having connectivity I can't play around with the drivers... Tomorrow I will try to set up a WLAN device and connect to the wireless network. Hopefully I gain connectivity and I can gain in-depth log knowledge!

Thanks again from me and my dad haha!

 

[jsterr]

No problem, sounds like a fun time with your dad ! Maybe take a look at this: https://www.taste-of-it.de/debian-i...tek-rtl8111-rtl8168-erhaelt-keine-ip-adresse/ could maybe be also used in pve (as it is based on debian bulllseye (pve7) or debian bookworm (pve8)) But linux-headers in pve need to be installed differently.

Edit: That might be more useful as it is the same problem that you have: https://www.reddit.com/r/Proxmox/co...=web3xcss&utm_term=1&utm_content=share_button

 

[techopsguy]

I tried everything, no success

I could never even have 10 minutes of link up, so couldn't play around the repos to download older versions or whatever. Many commands would fail giving me weird outputs (not found, permission denied...)

Ended up formatting the Proxmox partition and installed version 8. Even on the USB boot I could already get an IP assigned via dhcp...

Thanks again for your help man! My dad wants to compile his kernel by himself now haha

 

[Dunuin]

There is a whole thread about RTL8168/8169 not working with PVE8: https://forum.proxmox.com/threads/system-hanging-after-upgrade-nic-driver.129366/.

 

[techopsguy]

There is a whole thread about RTL8168/8169 not working with PVE8: https://forum.proxmox.com/threads/system-hanging-after-upgrade-nic-driver.129366/.

I've read the whole thing up and down. They have intermittent use of the NICs. They'd go up and after 1h, they go down and you have to reboot and stuff like that. In my case they were always down, nothing to make them come up. I also had a lot of commands not working due to a sloppy upgrade probably.