(Solved) In house cabling dropping

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

vrod

Active Member
Jan 18, 2015
241
43
28
31
Hello everybody,

I recently moved to a new place and of course one of my wishes was to have cabled connections from the cellar to all rooms, as I don't trust wireless (also because I'd like to run 10g in near future).

I pulled a cable, max 30 meter from my cave (mancave) to the cellar but I am getting some very weird iperf results. Sending data to my server (vm on esxi) gets me a speed of over 940mbps which makes sense, however when receiving, on both my 5960X desktop and laptop I achieve a max of ~430mbps.

I connected the desktop to a iscsi drive on the server as well and during big transfers the connection would suddenly drop and stay dead until i replugged the cable.

I'm a little at loss here, the only thing I believe it could be is that I used UTP cabling (which should be ok for homes). Any ideas?

Thanks,
Chris
 

fractal

Active Member
Jun 7, 2016
309
69
28
33
That kinda smells like an A/B wiring mismatch. Did you pull the wire yourself? Did you make sure to use the same pinout on both ends (A or B, but be the same for each pull)? Do both ends auto negotiate 1G full? It could also be a broken lead. They make 20 dollar testers that will tell you whether you have crossed wires or broken wires.
 

vrod

Active Member
Jan 18, 2015
241
43
28
31
Hello,

Thanks for your answer, both ends negotiate 1g and all wires checks out in correct order. I did B on both ends (half-orange, orange at start).
 

vrod

Active Member
Jan 18, 2015
241
43
28
31
OK so I have now tried to pull some SFTP (triple shielded) CAT6 cable to the top floor and it has the EXACT same issue! Negotiates 1Gbps and Sends with 1Gbps, however downloads with only 4-500Mbps..... I am at complete loss here, if there was too much interference, then the SFTP would do better anyway than the UTP cable, right?

I am open to any suggestions that might point somewhere. Completely confused right now. :(
 

ttabbal

Active Member
Mar 10, 2016
743
207
43
47
That depends... Shielding has to be "just so" to really be effective. Properly grounded with the proper connectors and such. If you want 10G anyway, why not just pull fiber? Zero interference, and 10G right now.

Note that you are testing with disks. That can present symptoms just due to overhead that have nothing to do with networking. Try something like iperf, test the raw network transfer. It could also be software related, the network stack in older Windows releases seems to have issues, though I only seem to remember issues on 10G. You could also have a bad NIC or connector on the NIC causing problems.

I would boot live Linux distributions on both ends and test using iperf. This does a couple things. It eliminates any overhead above the network layer. Allowing you to test what the network can do, not what the drive, controller, iscsi stack, etc. are doing. It also removes any configuration you might have done, or had done for you by Windows drivers, that might be causing problems. If that works, do a similar test with the normal OS booted up on one end. You have to eliminate possible failure points one at a time.
 

vrod

Active Member
Jan 18, 2015
241
43
28
31
Hello,

Thanks for your reply, I've been using iperf the entire time of course. :)

So there's fortunately and very surprisingly some good news... I think I have pinpointed the issue to actually be the laptop I was testing from!

I attached the laptop down in the celler to the switch directly and had the exactsame issue as on the 1st and 2nd floor.

I took another laptop (i use it to code vehicles), attached it and ran iperf. It gets 940Mbps both ways! Currently it's running since 5 minutes and there's no problems at all. Very surprised but also wondering how it could be. For instance, my 5960X machine had the same issue, however it also had some weird difficulties with the lc fiber cable I pulled in addition to the CAT6 cable... Maybe the system just needs a restart, could be that I've messed up with so many drivers on it. :D

However, I'll continue testing and eventually close the thread once I can confirm that the issues are no longer present.
 

Terry Kennedy

Well-Known Member
Jun 25, 2015
1,140
594
113
New York City
www.glaver.org
So there's fortunately and very surprisingly some good news... I think I have pinpointed the issue to actually be the laptop I was testing from!
A lot of laptops aren't really optimized for GigE. I have an older Dell Studio 1558 w/ a Realtek RTL8168D that has problems even at 500Mbit/sec.

Dell seems to have "solved" this - an Inspiron 17" from a year or 2 ago only supports 10/100, not GigE. :mad:
 

vrod

Active Member
Jan 18, 2015
241
43
28
31
Yeah I can agree on that, however my lenovo used to work pretty well when it came to transferring big files (backup, iso files and so on). Had been working on my desktop so just got it back up today. Still had the same issue, however this seemed to be driver related. Got the newest driver from ASUS (which seems to be a native MS driver lol) and it runs a stable 1 gig connection through iperf or SMB transfer. :)

[ 5] 284.00-285.00 sec 108 MBytes 903 Mbits/sec
[ 5] 285.00-286.00 sec 111 MBytes 933 Mbits/sec
[ 5] 286.00-287.00 sec 108 MBytes 907 Mbits/sec
[ 5] 287.00-288.00 sec 112 MBytes 937 Mbits/sec
[ 5] 288.00-289.00 sec 107 MBytes 895 Mbits/sec
[ 5] 289.00-290.00 sec 112 MBytes 936 Mbits/sec
[ 5] 290.00-291.00 sec 105 MBytes 881 Mbits/sec
[ 5] 291.00-292.00 sec 109 MBytes 918 Mbits/sec
[ 5] 292.00-293.00 sec 107 MBytes 896 Mbits/sec
[ 5] 293.00-294.00 sec 112 MBytes 937 Mbits/sec
[ 5] 294.00-295.00 sec 107 MBytes 898 Mbits/sec
[ 5] 295.00-296.00 sec 111 MBytes 928 Mbits/sec
[ 5] 296.00-297.00 sec 106 MBytes 892 Mbits/sec
[ 5] 297.00-298.00 sec 111 MBytes 935 Mbits/sec
[ 5] 298.00-299.00 sec 108 MBytes 904 Mbits/sec
[ 5] 299.00-300.00 sec 105 MBytes 880 Mbits/sec
[ 5] 300.00-300.04 sec 4.30 MBytes 940 Mbits/sec

Afterall a pretty strange issue with 2 "faulty" devices. Happy it's fixed now though :D