Not getting gigabit on my office ethernet

0

I've got a new fiber based internet connection and new gigabit router (ZyXEL Wireless VDSL2 Gateway) whereas my old router was only 100Mb/s.

My desktop computer is in my office in a 4 years old extension to the house where the electricians mounted an ethernet socket as well as in the utility room where router is placed. The cable running somewhere under the floor or in the walls (I don't know exactly) is a CAT6 of around 12 m.

With the new router I expected the network adapter to switch to gigabit mode, but it is still running 100Mb/s. The output from ethtool is shown below. If I instead pull a 15 years old and quite bad looking CAT5 cable of similar length over the floor, I get the full 1000Mb/s speed.

My thought is that there is something wrong with the cable they used or the sockets. When they made the extension I didn't have any gigabit equipment so I never thought of testing the speed.

1) Is it possible without special equipment to test the cables and sockets? I have various multimeters and a pocket LCD scope.

2) Could it be something else, maybe something obvious I didn't think of?

Output from ethtool:

Settings for eth1:
Supported ports: [ TP ]
Supported link modes:   10baseT/Half 10baseT/Full 
                        100baseT/Half 100baseT/Full 
                        1000baseT/Full 
Supported pause frame use: No
Supports auto-negotiation: Yes
Advertised link modes:  10baseT/Half 10baseT/Full 
                        100baseT/Half 100baseT/Full 
                        1000baseT/Full 
Advertised pause frame use: No
Advertised auto-negotiation: Yes
Speed: 100Mb/s
Duplex: Full
Port: Twisted Pair
PHYAD: 2
Transceiver: internal
Auto-negotiation: on
MDI-X: on (auto)
Cannot get wake-on-lan settings: Operation not permitted
Current message level: 0x00000007 (7)
               drv probe link
Link detected: yes
linux
ethernet
gigabit-ethernet
asked on Super User Jun 24, 2016 by marlar

3 Answers

2

Given the testing you have already done it most certainly sounds like your cable. If you can see the cable run I would look for

  • bad connection at the wall behind the plate. These cause slowdowns pretty regularly due to bad in pin outs. Make sure it's pinned out correctly too. This is the easiest and cheapest problem.
  • a splice in the cable, may be two segments spliced together somewhere in the middle you cant see. I have seen contractors splice bits and pieces of cable to save a trip to buy proper length.
  • cable running parallel to a power outlet service or any other power close by in the run. This causes loss. It should also not cross any lighting within 1m.

If you think you could pull in a new cable using the old one to pull it in I would do that. It obviously won't work if it's stapled down anywhere.

answered on Super User Jun 24, 2016 by (unknown user)
1

There are very specific tools to test cabling, they range from simple continuity testers for under $20 to highly advanced testing tools that costs thousands, even tens of thousands.

But here are multiple things to check...

1) Take the computer (or any computer) to the router and connect it directly to it, do you get gigabit?

1A) Take a computer/laptop with a known working gigbit connection and connect a patch cord directly between it and your computer and check the connection speed, it should be 1000.

3) Check patch cords are at least CAT-5e (CAT-6 is not necessary for Gigabit), verify they have all 8 conductors since it is (was) common especially for inexpensive switch/router manufacturers to include a 4-conductor patch cord. Patch cords with only pins 1, 2, 3, and 6 will only support 100mbps.

3) Check the jack terminations, this is common fault area, especially when done by people who are not properly trained or informed. Structured Wiring How To has a good page on general information on cable termination. This video from Ideal Tools shows how "professionals" do the job.

If you get to this point and everything looks fine, considering the tools you have available, I would do a simple continuity test, make sure each pin has continuity with each conductor having low resistance (14 ohms/100m is typical).

TBH, if after all this things look fine, my next step is to assess the difficulty in replacing the cable... If I can do it in less than an hour or two, I just do it. If it is a wiring "mess" and difficult to replace, I may go back to the shop and get the Fluke meter to see if I missed something obvious.

That being said, I have an HP Probook 650 laptop with Linux Mint and Windows 7 Pro on it, there are some switches out there that Mint will never establish a link to and Windows works fine, and I can't explain that, we have a couple Intellinet Gigabit switches in our test bench for testing devices and for some reason when I boot into Mint it will not connect to the switch and get an IP, but it works fine in Windows but if I connect to one of our infrastructure HP Procurve switches it works fine.

answered on Super User Jun 24, 2016 by acejavelin
0

Just a heads-up on this one. I myself ran into this with various cables and such. I found that I could set it manually to gigabit with ethtool but it would reset in a few seconds. Then I checked my tuned profile. If you're using an energy saving tuned-adm profile such as 'powersave' it may be dropping network speed for you. Setting back to something like "network-throughput" immediately kicks in high gear at 1000mb.

Rather than put the whole system in a higher power state, I resolved to create a custom tuning profile from 'powersave' with only the wifi set to powersave:

# /usr/lib/tuned/powersave/tuned.conf
# tuned configuration
#

[main]
summary=Optimize for low power consumption

[cpu]
governor=ondemand
energy_perf_bias=powersave

[eeepc_she]

[vm]

[audio]
timeout=10

[video]
radeon_powersave=auto

[disk]
# Comma separated list of devices, all devices if commented out.
# devices=sda

[net]
# Comma separated list of devices, all devices if commented out. (change this)
devices=wlp1s0

[scsi_host]
alpm=min_power

[sysctl]
vm.laptop_mode=5
vm.dirty_writeback_centisecs=1500
kernel.nmi_watchdog=0

[script]
script=${i:PROFILE_DIR}/script.sh

Maybe it's a bonehead situation, but hope this helps out someone down the road. Thx

answered on Super User May 26, 2017 by BoeroBoy

User contributions licensed under CC BY-SA 3.0