Proxmox 8.3.5 and Connection Errors

[spetrillo]

Hello all,

I have Proxmox 8.3.5 running on an 8th gen NUC. After upgrading to 8.3.5 I am noticing that I am getting intermittent connection errors when viewing a VM running on this server. Its like communication to Proxmox is lost and then it recovers. I am not seeing this on any other 8.3.5 servers. Is anyone else seeing this?

Thanks,
Steve

 

[dakralex]

Hi!

If it happens only on one server, then it'd be most likely the fault of the connection to the server. How is the network setup? What is the performance of the link between your host and the Proxmox VE server? Is there any packet loss?

 

[anconakip]

i have the same sort problem. After update to 8.3.5 my proxmox service loses connection after some time. Help!

 

[dakralex]

i have the same sort problem. After update to 8.3.5 my proxmox service loses connection after some time. Help!

Could you provide more information what exactly does not work? Are there any errors in the syslog? What does happen if a local and an external IP/domain is pinged?

 

[anconakip]

Could you provide more information what exactly does not work? Are there any errors in the syslog? What does happen if a local and an external IP/domain is pinged

Could you provide more information what exactly does not work? Are there any errors in the syslog? What does happen if a local and an external IP/domain is pinged?

Below the problem noted in the syslog.

Apr 08 02:40:26 pve kernel: e1000e 0000:00:19.0 enp0s25: Detected Hardware Unit Hang: TDH <66> TDT <89> next_to_use <89> next_to_clean <65>buffer_info[next_to_clean]: time_stamp <101554bc2> next_to_watch <66> jiffies <101556d40> next_to_watch.status <0>MAC Status <80083>PHY Status <796d>PHY 1000BASE-T Status <3800>PHY Extended Status <3000>PCI Status <10>

 

[dakralex]

Below the problem noted in the syslog.

Did the upgrade also include a proxmox-kernel upgrade? Which kernel version are you running on (you can check with uname -v)?

Otherwise, the log seems to indicate that there was some transmission task that did not complete in time. This is usually caused by the on-chip LAN controller offloading some network transmission and not being able to keep up with the load. This can usually be mitigated by turning off some of those acceleration features. The usual candidates (RX/TX checksumming, TCP segmentation offload) are turned off with this command, but you can go one-by-one to check which seems to cause the problem:

ethtool -K <dev> rx off tx off tso off

This can also be added to the interface config as a post-up hook, so it is set at boot time.

 

[anconakip]

Did the upgrade also include a proxmox-kernel upgrade? Which kernel version are you running on (you can check with uname -v)?

Otherwise, the log seems to indicate that there was some transmission task that did not complete in time. This is usually caused by the on-chip LAN controller offloading some network transmission and not being able to keep up with the load. This can usually be mitigated by turning off some of those acceleration features. The usual candidates (RX/TX checksumming, TCP segmentation offload) are turned off with this command, but you can go one-by-one to check which seems to cause the problem:

ethtool -K <dev> rx off tx off tso off

This can also be added to the interface config as a post-up hook, so it is set at boot time.

thanks. I should try this.

this is the output of uname -v: #1 SMP PREEMPT_DYNAMIC PMX 6.8.12-9 (2025-03-16T19:18Z)

 

[Nico_DreamzZ]

I have also problems with my intel nuc (Intel NUC 10I5FNHN2) with proxmox 8.4.1 after some hours and sometimes days the network connection dropps, after a replug of the ethernet cable it works again, a reboot also does fix the situation. Kernel is: Linux pve 6.8.12-9-pve #1 SMP PREEMPT_DYNAMIC PMX 6.8.12-9 (2025-03-16T19:18Z) x86_64 GNU/Linux

I just entered the

ethtool -K eno1 rx off tx off tso off

manually, hope this fixes the situation.