Featured image of post Proxmox Stability Fix: Intel I219-LM/V NICs
Tutorials

Proxmox Stability Fix: Intel I219-LM/V NICs

Proxmox Stability Fix: Intel I219-LM/V NICs

This tutorial provides a permanent solution to prevent complete network link failures on Proxmox VE hosts using the Intel I219-LM or I219-V integrated Network Interface Cards (NICs), commonly found in devices like the Lenovo ThinkCentre M720q Tiny nodes.

These NICs have a critical bug that causes the network link to hang completely when used with Linux bridges (vmbr0) and hardware offloading enabled. The interface becomes unresponsive, requiring a full system reboot to restore connectivity. Disabling Transmit and Receive Checksum Offloading via ethtool prevents these catastrophic failures.

The Problem

Users typically experience:

  • Complete loss of network connectivity without warning
  • Network interface becomes totally unresponsive
  • Often occurs under moderate to heavy network load
  • Affects both host and all VMs/containers using the NIC

Prerequisites

  • Administrative SSH access to your Proxmox VE host
  • The name of your physical network interface (typically eno1, enp1s0, or similar)
  • Basic familiarity with command-line text editors (nano, vim, etc.)

Step 1: Identify Your Network Interface and Current Settings

First, confirm the name of your Intel I219-LM/V interface and check its current offloading status.

1
2
3
4
5

# 1. Identify your NIC and its logical name
sudo lshw -C network | grep -E 'product|logical name'

# 2. Check current offloading status (replace eno1 with your interface name)
ethtool -k eno1 | grep "checksumming"

Expected output: You’ll likely see on for both tx-checksumming and rx-checksumming, indicating offloading is currently enabled (the problematic state).

Step 2: Edit the Proxmox Network Configuration

Proxmox VE uses the standard Debian networking configuration file (/etc/network/interfaces) based on the ifupdown system.

  1. Create a backup of your current configuration:

    1

    sudo cp /etc/network/interfaces /etc/network/interfaces.backup

  2. Open the configuration file for editing:

    1

    sudo nano /etc/network/interfaces

  3. Locate the configuration block for your primary management bridge (typically vmbr0) where your physical interface is attached as a bridge port.

Step 3: Apply the Permanent Fix Using post-up

Add a post-up command to the bridge configuration. This command executes after the bridge is fully initialized, disabling offloading before any network traffic can trigger the hang.

Modify your vmbr0 configuration block as follows:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11

# /etc/network/interfaces
auto vmbr0
iface vmbr0 inet static
    address 192.168.1.10/24
    gateway 192.168.1.1
    bridge-ports eno1          # Your physical interface
    bridge-stp off
    bridge-fd 0
    # --- CRITICAL FIX: PREVENT INTEL NIC HANG ---
    post-up /usr/sbin/ethtool -K eno1 tx off rx off
    # ---------------------------------------------

Important: Replace eno1 with your actual interface name identified in Step 1.

  1. Save the file and exit the editor (in nano: Ctrl+X, then Y, then Enter).

Step 4: Apply Changes and Verify

Since these changes affect the core network stack and you want to ensure they’re applied before the bug can trigger, a reboot is strongly recommended.

  1. Reboot the Proxmox host:

    1

    sudo reboot

  2. After reboot, verify the settings via SSH:

    1

    ethtool -k eno1 | grep "checksumming"

    Expected output (confirming successful fix):

    1
2

    rx-checksumming: off
tx-checksumming: off

Alternative: Apply Without Reboot (Use with Caution)

If you cannot reboot immediately and the network is currently working:

1
2
3
4
5

# Apply settings immediately to prevent future hangs
sudo ethtool -K eno1 tx off rx off

# Restart networking (may briefly interrupt connections)
sudo systemctl restart networking

Conclusion

Your Lenovo M720q Tiny node (or similar hardware with Intel I219-LM/V NICs) should now run reliably without experiencing sudden, complete network failures. This fix prevents:

  • Complete network link hangs requiring hard reboots
  • Loss of access to the Proxmox host
  • Disruption to all VMs and containers
  • Unexpected downtime in production environments

This solution has been extensively tested and confirmed to completely eliminate the network hang issue on affected hardware.