Weird cabling issue (Solved)

[ColinTaylor]

Glad you found the problem. Murphy's law means that whenever you change your network something else will change at the same time just to confuse you.

Without wishing to dwell on it, this was a reason I asked you this question earlier: "And this is the same PC that you used to do the test when connected directly into the router on the 1st floor".

 

[Samir]

Nice! Glad you've figure it out.

When isolating cables for testing, the best approach is the cable and two systems running iperf--one of the most robust cable tests around.

 

[ColinTaylor]

When isolating cables for testing, the best approach is the cable and two systems running iperf--one of the most robust cable tests around.

And that's exactly what he did near the beginning of this thread.

 

[Samir]

And that's exactly what he did near the beginning of this thread.

I saw that, but the results were not conclusive since the bandwidth was only good in one direction.

I think there was an issue with the keystone that was corrected, but then the test was not repeated in the same manner, otherwise I think it would have been full speed in both directions proving the cable was fine.

 

[coxhaus]

I did networking for many years and if it was a cable issue you would see it in the stats.

The hardest problem in the old days was when someone put a Ethernet 50 ohm cable terminator on a Arcnet 75 ohm network cable. It caused weird reflections on the network. Things are a lot easier using CAT5e cable now days.

 

[TheLostSwede]

Glad you found the problem. Murphy's law means that whenever you change your network something else will change at the same time just to confuse you.

Without wishing to dwell on it, this was a reason I asked you this question earlier: "And this is the same PC that you used to do the test when connected directly into the router on the 1st floor".

Missed that part, but no, as this is a desktop and the one on the first floor was a laptop...

 

[TheLostSwede]

Nice! Glad you've figure it out.

When isolating cables for testing, the best approach is the cable and two systems running iperf--one of the most robust cable tests around.

The thing is, I was getting seemingly slower iPerf numbers as well, most likely again thanks to the "QoS" software
I had completely forgotten about it and it didn't do any weird stuff until the recent version.
It was just something I was testing out, but it turns out it was a bad thing to play around with...

 

[TheLostSwede]

I did networking for many years and if it was a cable issue you would see it in the stats.

The hardest problem in the old days was when someone put a Ethernet 50 ohm cable terminator on a Arcnet 75 ohm network cable. It caused weird reflections on the network. Things are a lot easier using CAT5e cable now days.

Ah, the days of BNC networks... Had that in school.
Friend of mine with rich parents at the time set up a token ring network at home and he had a 100m cable so he could use his laptop in the garden...
Admittedly the guy started one of the first ISPs in Sweden a couple of years later, but even so...

 

[TheLostSwede]

Looks like it's all good now after a Windows update on top of getting rid of the software.

 

[Smokindog]

And don't forget cleaning up your connections, fixing your punch downs and gaining valuable technique Now go buy some basic tools before doing that again )

Glad you're up and running. It was a pleasure chatting with you!

 

[Samir]

I set up our first networks with 10Base-2. That whole terminating thing was interesting, but no different than SCSI at the time. What was nice was the distance you could run with rg wire. I actually still have that wire, the Ts and the terminators, and the $900 switches we bought back then--all still working. (They better be for how expensive they were!)

 

[coxhaus]

Ah, the days of BNC networks... Had that in school.
Friend of mine with rich parents at the time set up a token ring network at home and he had a 100m cable so he could use his laptop in the garden...
Admittedly the guy started one of the first ISPs in Sweden a couple of years later, but even so...

Yes token ring. I have an IBM 8 port MAU in my attic.

 

[coxhaus]

I set up our first networks with 10Base-2. That whole terminating thing was interesting, but no different than SCSI at the time. What was nice was the distance you could run with rg wire. I actually still have that wire, the Ts and the terminators, and the $900 switches we bought back then--all still working. (They better be for how expensive they were!)

Yes 10base-2, Arcnet was longer range distance wise but lower bandwidth. We used Arcnet because we had 20 campuses with lots of buildings which 10base-2 could not handle. We finally replaced all of it with fiber from building to building and CAT5e inside buildings.

 

[Samir]

Yes token ring. I have an IBM 8 port MAU in my attic.

I have 3 in the 'computer room'

What was interesting about Token Ring was that under Lantastic I could bridge Lantastic across Token Ring and Ethernet if I had both cards in a system. I was going to run dual networks in all my systems for extra bandwidth. This was pre-Internet, so things changed a bit...

 

[Samir]

Yes 10base-2, Arcnet was longer range distance wise but lower bandwidth. We used Arcnet because we had 20 campuses with lots of buildings which 10base-2 could not handle. We finally replaced all of it with fiber from building to building and CAT5e inside buildings.

Yep, that 2Mb limit from what I recall. I never really understood why Arcnet was still around when 10Base-2 was there, but you'd still see the hardware all the time--probably because it worked on things Ethernet didn't at the time.

Nice jump from the coax to the fibre!

 

[coxhaus]

I forgot we did have a mile long fiber token ring in the real old days on a campus. 16 meg token ring was faster than 10 meg ethernet hub. There were no switches back then. Plus token ring ran real well loaded down with traffic. Ethernet hubs bogged down when they were loaded.

 

[degrub]

we beta tested the first fiber optic token ring modems in our plant and helped the manufacturer get them to work.

 

[Samir]

16 meg token ring was faster than 10 meg ethernet hub. There were no switches back then. Plus token ring ran real well loaded down with traffic. Ethernet hubs bogged down when they were loaded.

Oh yeah, that's one of the reasons I wanted to get token ring working to compare the two. But switches came along and so did Fast Ethernet and token ring got left behind and then so did atm.

 

[coxhaus]

I have now been thinking about old networks. My first home network was token ring. It was because I wanted a networked printer and I came across a token ring printer server. Back then you needed an expensive printer server box to plug your parallel printer port into. I had bought an expensive postscript laser printer at Best Buy. I had to pay it out in installments. So token ring was my home topology. I had a 286 PC and a 486 PC running OS/2.

 

[Samir]

I have now been thinking about old networks. My first home network was token ring. It was because I wanted a networked printer and I came across a token ring printer server. Back then you needed an expensive printer server box to plug your parallel printer port into. I had bought an expensive postscript laser printer at Best Buy. I had to pay it out in installments. So token ring was my home topology. I had a 286 PC and a 486 PC running OS/2.

That was a super-advanced setup for the time!

It took a while before we got all networked and it was ethernet cards and 10Base-2 at the beginning. But I was smart enough to by combo nics so when we later moved certain topologies to 10Base-T, we were able to just plug in. The switch also had a combo port (including an aui), as well as 2 fast ethernet uplinks to make our 'backbone'. Still have those switches almost 25 years later.