Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.debian.user > #222469 > unrolled thread

1Gbps Ethernet drops to 100Mbps

Back to article view | Back to linux.debian.user

Contents

  1Gbps Ethernet drops to 100Mbps Nazar Zhuk <nazar@zhuk.online> - 2020-05-13 21:00 +0200
    Re: 1Gbps Ethernet drops to 100Mbps Andy Smith <andy@strugglers.net> - 2020-05-14 08:00 +0200
    Re: 1Gbps Ethernet drops to 100Mbps Michael Stone <mstone@debian.org> - 2020-05-14 14:00 +0200
    Re: 1Gbps Ethernet drops to 100Mbps Nazar Zhuk <nazar@zhuk.online> - 2020-05-19 21:40 +0200

#222469 — 1Gbps Ethernet drops to 100Mbps

Hi all,

I have a 1Gbps network port that correctly connects as 1Gbps full duplex 
on boot, then drops to 100Mbps 4 seconds later.

Forcing 1Gbps full duplex with ethtool works fine. Tested to make sure 
the actual speed is over 100Mbps.

I am trying to determine what causes the drop to 100Mbps and how to 
avoid it.

The network port is on a Thunderbolt 3 hub in case this matters. Though 
it appears that it looks just like a regular 1Gbps card to igb driver.

Tried two different Cat6 cables at 5 ft distance to the switch.

There is nothing aside from loopback in /etc/network/interfaces and 
/etc/network/interfaces.d.


Relevant log entries:

# dmesg | grep -E '(igb|eth|enp9s0)'
[    1.707669] wmi_bus wmi_bus-PNP0C14:01: WQBC data block query control 
method not found
[    5.578360] igb: Intel(R) Gigabit Ethernet Network Driver - version 
5.4.0-k
[    5.578360] igb: Copyright (c) 2007-2014 Intel Corporation.
[    6.351902] igb 0000:09:00.0: added PHC on eth0
[    6.351903] igb 0000:09:00.0: Intel(R) Gigabit Ethernet Network 
Connection
[    6.351904] igb 0000:09:00.0: eth0: (PCIe:2.5Gb/s:Width x1) 
64:4b:f0:10:5e:80
[    6.352004] igb 0000:09:00.0: eth0: PBA No: 000300-000
[    6.352005] igb 0000:09:00.0: Using MSI-X interrupts. 4 rx queue(s), 
4 tx queue(s)
[    6.352464] igb 0000:09:00.0 enp9s0: renamed from eth0
[   71.234244] IPv6: ADDRCONF(NETDEV_UP): enp9s0: link is not ready
[   71.262391] IPv6: ADDRCONF(NETDEV_UP): enp9s0: link is not ready
[   74.272257] igb 0000:09:00.0 enp9s0: igb: enp9s0 NIC Link is Up 1000 
Mbps Full Duplex, Flow Control: RX/TX
[   74.487967] IPv6: ADDRCONF(NETDEV_CHANGE): enp9s0: link becomes ready
[   78.623864] igb 0000:09:00.0 enp9s0: igb: enp9s0 NIC Link is Up 100 
Mbps Half Duplex, Flow Control: None
[   78.623869] igb 0000:09:00.0: EEE Disabled: unsupported at half 
duplex. Re-enable using ethtool when at full duplex.


Then forcing with ehttool:

# ethtool -s enp9s0 speed 1000 duplex full

# dmesg | grep -E '(igb|eth|enp9s0)'
...
[  346.656896] igb 0000:09:00.0 enp9s0: igb: enp9s0 NIC Link is Up 1000 
Mbps Full Duplex, Flow Control: RX/TX


Any ideas would be much appreciated. Thanks.

-- 
Nazar

[toc] | [next] | [standalone]

#222477

Hi Nazar,

On Wed, May 13, 2020 at 08:58:16PM +0200, Nazar Zhuk wrote:
> I have a 1Gbps network port that correctly connects as 1Gbps full duplex on
> boot, then drops to 100Mbps 4 seconds later.

Negotiation is failing and that is required for 1Gbit, so I would
start by ruling out the device that your computer is plugged into by
plugging it into something else.

A real gigabit switch would be ideal but even another gigabit port
on a computer will work well enough to show negotiation, as
auto-crossover (Auto MDI-X) is implemented on all but the most basic
gigabit chips.

If it works with other things then the conclusion is probably that
your Thunderbolt hub is junk. If it doesn't work with other things
then the conclusion is probably that your network chipset is junk.
It would be very very unlikely to be a software issue in Debian or
the kernel as negotiation is at a very low level. But if it does not
work properly with other things you could try reporting it as an
upstream kernel bug in the driver and see what happens.

Cheers,
Andy

-- 
https://bitfolk.com/ -- No-nonsense VPS hosting

[toc] | [prev] | [next] | [standalone]

#222493

On Wed, May 13, 2020 at 08:58:16PM +0200, Nazar Zhuk wrote:
>I have a 1Gbps network port that correctly connects as 1Gbps full 
>duplex on boot, then drops to 100Mbps 4 seconds later.

I'd check that autonegotiation isn't disabled on the other end, then try 
another cable.

[toc] | [prev] | [next] | [standalone]

#222652

On 5/13/20 1:58 PM, Nazar Zhuk wrote:
> 
> I have a 1Gbps network port that correctly connects as 1Gbps full duplex 
> on boot, then drops to 100Mbps 4 seconds later.
> 
> Forcing 1Gbps full duplex with ethtool works fine. Tested to make sure 
> the actual speed is over 100Mbps.
> 
> I am trying to determine what causes the drop to 100Mbps and how to 
> avoid it.

Thanks for the hardware suggestions. It's not that. For whatever reason 
NetworkManager was configured to force 100Mbps. It's a new install and I 
don't remember doing that, but that's what it was.

-- 
Nazar

[toc] | [prev] | [standalone]

Back to top | Article view | linux.debian.user

csiph-web