r8169 NETDEV WATCHDOG Timeout Causing Network Drop

[amfzn]

Hello, I need suggestion which commands to try in order to find the solution to the network issue:

## Problem Summary
Faced network interface drop with r8169 driver causing "NETDEV WATCHDOG" timeout errors, eventually dropping the link (Link is Down/Up cycle), no LAN or router connectivity.

journal displayed:

## Log Evidence

Jul 07 21:07:29 kernel: r8169 0000:03:00.1 eno1: NETDEV WATCHDOG: CPU: 0: transmit queue 0 timed out 5001 ms
Jul 07 21:07:29 kernel: r8169 0000:03:00.1: can't disable ASPM; OS doesn't have ASPM control
Jul 07 21:07:29 kernel: r8169 0000:03:00.1 eno1: rtl_txcfg_empty_cond == 0 (loop: 42, delay: 100).
Jul 07 21:07:29 kernel: r8169 0000:03:00.1 eno1: rtl_rxtx_empty_cond == 0 (loop: 42, delay: 100).
Jul 07 21:07:35 kernel: r8169 0000:03:00.1 eno1: NETDEV WATCHDOG: CPU: 4: transmit queue 0 timed out 5305 ms
Jul 07 21:07:35 kernel: r8169 0000:03:00.1: can't disable ASPM; OS doesn't have ASPM control
Jul 07 21:07:35 kernel: r8169 0000:03:00.1 eno1: rtl_txcfg_empty_cond == 0 (loop: 42, delay: 100).
Jul 07 21:07:35 kernel: r8169 0000:03:00.1 eno1: rtl_rxtx_empty_cond == 0 (loop: 42, delay: 100).
Jul 07 21:07:40 kernel: r8169 0000:03:00.1 eno1: NETDEV WATCHDOG: CPU: 4: transmit queue 0 timed out 5568 ms
Jul 07 21:07:40 kernel: r8169 0000:03:00.1: can't disable ASPM; OS doesn't have ASPM control
Jul 07 21:07:40 kernel: r8169 0000:03:00.1 eno1: rtl_txcfg_empty_cond == 0 (loop: 42, delay: 100).
Jul 07 21:07:40 kernel: r8169 0000:03:00.1 eno1: rtl_rxtx_empty_cond == 0 (loop: 42, delay: 100).


$ sudo lspci -vvv -s $(lspci | grep Ethernet | cut -d' ' -f1) | grep -i power
Capabilities: [40] Power Management version 3
ExtTag+ AttnBtn- AttnInd- PwrInd- RBE+ FLReset- SlotPowerLimit 26W
EmergencyPowerReduction Not Supported, EmergencyPowerReductionInit-

$ apt list r8168-dkms -a
r8168-dkms/noble-updates 8.052.01-1ubuntu1.1 all
r8168-dkms/noble 8.052.01-1ubuntu1 all

Driver Info

driver: r8169
version: 6.17.0-35-generic
bus-info: 0000:03:00.1

Root Cause Analysis (AI)

The Realtek r8169 driver exhibits known instability when ASPM (Active State Power Management) cannot be properly controlled by the OS, leading to transmit queue timeouts and eventual link failure. This affects many users with compatible NIC hardware.

Immediate Workarounds (No Reboot Required)

Option 1: Disable Link Power Management via ethtool (Temporary)

sudo ethtool -s eno1 wol d
sudo ethtool -K eno1 gso off tso off

Effect: Disables Wake-on-LAN power states and large offloads that may trigger timeouts. May reduce performance slightly.

Option 2: Reload r8169 Module Live (Risk: Brief Disconnection)

sudo modprobe -r r8169 && sudo modprobe r8169 ASPM=0

Effect: Forces module reload with ASPM disabled. Requires root, brief network interruption during reload.

Option 3: Monitor and Auto-Recover via Script

while true; do
ping -c1 192.168.1.1 >/dev/null 2>&1 || sudo ip link set eno1 down && sleep 2 && sudo ip link set eno1 up;
sleep 10;
done

Effect: Automatic recovery if gateway unreachable. Run in background or systemd service.

Permanent Fix (Requires Reboot)

Users needing stability should consider:

Add kernel parameter pcie_aspm=off to GRUB
Blacklist r8169, use r8168-dkms driver instead
Update BIOS/firmware for improved PCIe power management

Verification

Post-fix check:

dmesg | grep -c "NETDEV WATCHDOG"

(Should return 0 after fix)

Notes

r8169 is open-source; some Realtek chipsets work better with proprietary r8168 driver
Check cable integrity and switch port as secondary troubleshooting
Report results to help community identify hardware/driver combinations

System: KDE Neon (Ubuntu 24.04 LTS), Kernel 6.17, Realtek PCIe GbE Controller (r8169)

Mainly which commands are safe to run and look like a possible solution to the above described issue?