OMV 7 - Persistent Network Issue: "e1000e detected hardware unit hang"

I had the same problem before virtualizing OMV on Proxmox; to solve it, you can create a scheduled task set as "at reboot" with the command you found.

Out of interest, what kernel are you using using? Stable 6.1 series , or backport 6.12 series?

Quote from Krisbee

Out of interest, what kernel are you using using? Stable 6.1 series , or backport 6.12 series?

Linux 6.14.8-3-bpo12-pve

Quote from 2Psycho

Hello everyone,

I'm an experienced OpenMediaVault user, having run OMV 6 and older versions successfully for a long time. I recently upgraded to OMV 7 (currently running the latest version) and have encountered a persistent and frustrating issue with my network interface that leads to frequent system freezes.

The Problem

Since the upgrade to OMV 7, my system intermittently reports the error and subsequently hangs:

Code

e1000e detected hardware unit hang

This is related to the e1000e driver, which is commonly used for Intel network cards. This is an very old driver issue which I expected to be fixed

The Temporary Fix

I have found a temporary workaround that resolves the issue immediately after a hang, but it is not persistent across reboots or interface restarts. By connecting via SSH (if the system hasn't completely frozen) or through a console, the following command restores stability:

Code

sudo ethtool -K eno1 gso off gro off tso off tx off rx off

(Note: I am using eno1 as my interface name, which may need to be adjusted by other users.)

This command disables various offloading features (Generic Segmentation Offload, Generic Receive Offload, TCP Segmentation Offload, Transmit Offload, and Receive Offload) that seem to be causing the instability with my specific hardware and the e1000e driver in the OMV 7 environment.

The Request for a Permanent Solution

My core request is to find a way to make these ethtool settings permanent.

1. OMV Interface Integration: Is there currently an option within the OMV 7 Web Interface (under System → Network → Interfaces, or elsewhere) to apply custom ethtool parameters to a specific network adapter? I have looked but may be overlooking where this functionality is hidden.
2. Permanent Configuration Method: If not, could the development team consider adding a field in the network interface settings to enter such startup parameters? This would greatly benefit users experiencing similar driver/hardware conflicts and prevent the need for manual workarounds or non-standard configuration file modifications.
3. Alternative: Is there a supported OMV/Debian method (like a specific configuration file or a different initialization script) that I can use to reliably apply these ethtool settings during the network service startup, which won't be overwritten by OMV's configuration management?

lamborghini urus for rent dubai

Any guidance on a supported, persistent fix, or an update on whether such a configuration option is planned, would be greatly appreciated. Thank you for your continued work on OpenMediaVault!

Best regards,

Display More

Yeah, that’s a nasty old bug. Your workaround makes sense though — disabling those offloads usually calms the e1000e driver down. You could make it persistent with a small systemd service or a startup script; that’s what I did on my Debian box and it’s been solid since.