I can confirm that this bug exists on 18.04 and 18.10 on a Intel NUC 8i3BEH (with latest BIOS, 0056). After a few seconds after boot, the RTT spikes to well over 1s and makes the system unusable.
Device:
00:1f.6 Ethernet controller: Intel Corporation Ethernet Connection (6) I219-V (rev 30)
Subsystem: Intel Corporation Ethernet Connection (6) I219-V
Flags: bus master, fast devsel, latency 0, IRQ 127
Memory at c0100000 (32-bit, non-prefetchable) [size=128K]
Capabilities: [c8] Power Management version 3
Capabilities: [d0] MSI: Enable+ Count=1/1 Maskable- 64bit+
Kernel driver in use: e1000e
Kernel modules: e1000e
Forcing the speed of the interface to 100 with autonegotiation off (via ethtool --change eno1 speed 100 autoneg off) helps somewhat - answer times stay around a few ms.
Since changing the power management settings seems to solve this issue, i tried various options: Disable intel_pstate, change the governor to performance, forcing full frequency, etc. without luck.
However, disabling PCIe ASPM *in the BIOS* solved the issue for me - at the price of a somewhat higher power consumption (around +5W).
Please note that using pcie_aspm=off in the kernel command line (linux-image-4.18.0-14-generic) *did not work*.
Here is the dmesg snippet with ASPM disabled in the BIOS:
# dmesg | grep -i aspm
[ 0.041040] ACPI FADT declares the system doesn't support PCIe ASPM, so disable it
[ 0.177567] acpi PNP0A08:00: _OSC: OS supports [ExtendedConfig ASPM ClockPM Segments MSI]
[ 0.181078] acpi PNP0A08:00: FADT indicates ASPM is unsupported, using BIOS configuration
I can confirm that this bug exists on 18.04 and 18.10 on a Intel NUC 8i3BEH (with latest BIOS, 0056). After a few seconds after boot, the RTT spikes to well over 1s and makes the system unusable.
Device:
00:1f.6 Ethernet controller: Intel Corporation Ethernet Connection (6) I219-V (rev 30)
Subsystem: Intel Corporation Ethernet Connection (6) I219-V
Flags: bus master, fast devsel, latency 0, IRQ 127
Memory at c0100000 (32-bit, non-prefetchable) [size=128K]
Capabilities: [c8] Power Management version 3
Capabilities: [d0] MSI: Enable+ Count=1/1 Maskable- 64bit+
Kernel driver in use: e1000e
Kernel modules: e1000e
# ethtool -i eno1 rom-version: statistics: yes eeprom- access: yes register- dump: yes priv-flags: no
driver: e1000e
version: 3.2.6-k
firmware-version: 0.4-4
expansion-
bus-info: 0000:00:1f.6
supports-
supports-test: yes
supports-
supports-
supports-
Forcing the speed of the interface to 100 with autonegotiation off (via ethtool --change eno1 speed 100 autoneg off) helps somewhat - answer times stay around a few ms.
Since changing the power management settings seems to solve this issue, i tried various options: Disable intel_pstate, change the governor to performance, forcing full frequency, etc. without luck.
However, disabling PCIe ASPM *in the BIOS* solved the issue for me - at the price of a somewhat higher power consumption (around +5W). image-4. 18.0-14- generic) *did not work*.
Please note that using pcie_aspm=off in the kernel command line (linux-
Here is the dmesg snippet with ASPM disabled in the BIOS:
# dmesg | grep -i aspm
[ 0.041040] ACPI FADT declares the system doesn't support PCIe ASPM, so disable it
[ 0.177567] acpi PNP0A08:00: _OSC: OS supports [ExtendedConfig ASPM ClockPM Segments MSI]
[ 0.181078] acpi PNP0A08:00: FADT indicates ASPM is unsupported, using BIOS configuration
Please let me know, if you need more information.