Networking Question on Long Ethernet Cables

Hey friends, I'm writing because I'm trying to make my connection to my PC better. When I put this all together years ago, when we first moved into this apartment, I had no idea there were different types of ethernet cables, but now I do. Basically setup is:

Cable Internet
|
v
Cable Modem
|
v
Netgear Nighthawk Router
|
v
Netgear Switch
|
v
Long cat5e cable snaked along the base boards
|
v
Cheap ethernet coupler
|
V
Long cat6 connected to the PC.

I get a decent amount of dropped frames and congestion during my streams, and I'm wondering if there's anything I can do to strengthen the connection between these long ethernet cables (wifi isn't an option).

So I'm asking for your help. Might y'all have an idea as to what I can do to make this better?

Thanks in advance for your help.

Having a switch on each end of a long run is something I'd recommend. Small and cheap is fine, as long as it's powered.

Hmmmm, so a second switch where the PC is? Like this:

https://www.amazon.com/Ethernet-Spli...

Veloxi wrote:

Hmmmm, so a second switch where the PC is? Like this:

https://www.amazon.com/Ethernet-Spli...

Yes. Though Cat5e and Cat6 should do fine for lengths of 100m+ between switches, long runs out to a single machine can be problematic without a switch on both ends. (It's one of those "shouldn't do this but sometimes does" things.) Other than that, going between cable types will have no impact as long as the runs are within each cable's spec for the speed you're operating at.

That's how I'd do a network regardless though. Your problems could still be due to a cable which has been damaged in some way. Enough to introduce come crosstalk which wouldn't interfere with everyday traffic to any noticeable degree, but may interfere when the link is saturated. If it doesn't go away with a switch there, that's what I'd look at.

Okay, so to confirm:

Original Switch -> Long Cable -> Coupler -> Long Cable -> New Switch -> PC

?

I've also been pinging my router and such all day, and am getting stuff like:

Reply from 8.8.8.8: bytes=68 (sent 1472) time=15ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=16ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=17ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=24ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=27ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=107ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=18ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=25ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=16ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=60ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=14ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=17ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=15ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=16ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=18ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=16ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=19ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=15ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=13ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=14ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=15ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=14ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=15ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=18ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=18ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=15ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=15ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=18ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=15ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=19ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=15ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=26ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=15ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=15ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=18ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=16ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=24ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=16ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=34ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=16ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=130ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=16ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=47ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=13ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=14ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=17ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=16ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=20ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=14ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=14ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=16ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=16ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=17ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=15ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=14ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=13ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=15ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=17ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=19ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=26ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=16ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=17ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=15ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=17ms TTL=54
Reply from 8.8.8.8: bytes=68 (sent 1472) time=16ms TTL=54

Ping statistics for 8.8.8.8:
Packets: Sent = 65, Received = 65, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 13ms, Maximum = 130ms, Average = 21ms

Does that mean things are mostly okay?

That IP you are pinging is a Google DNS server not your router. If that is your router IP then your configuration is super weird.

OMG, someone on another forum pointed out that certain modems have an intel chipset that could be problematic:

https://approvedmodemlist.com/intel-...

Guess who has one of those modems?! THIS GUY.

Try pinging the actual router; it will usually be the .1 in your internal network range. That is, if your PC has the IP address A.B.C.D, D will usually be around 50 or around 100, and the router will usually be on A.B.C.1.

You can see what your PC's gateway is by typing 'ipconfig' at a prompt. The IPv4 address is your PC, one of the first lines, and a couple lines below that will be Default Gateway. Unless you have a extremely odd setup, that will be your router. Ping that for awhile and see if it seems solid.

My guess is that you're probably fine internally. Ethernet is pretty resilient, and while I think you're slightly out of spec by having that many unpowered hops, it's probably not bad enough to cause noticeable frame loss. But running ping for a good long while will help prove that.

If you have a Linux box on your network, you can pretty easily look at wire error counts. I don't remember any easy way to get that info in Windows. I'm sure it must be available on some layer of the networking stack, but I don't remember how to dig it out.

If you do have a Linux, log into it and type 'ifconfig'. Look for the network card. It used to always be eth0, but modern Linuxes call the network card damn near anything; I think it's ens33 when I have Linux booted on this machine, for instance, where it's enp0s25 on another box.

In the network card's block of stats, there are two lines that are interesting in this context: RX errors and TX errors. Ideally, those should all be zeroes.

You can have a few errors without a problem. Oddly, for instance, I'm showing 2463 "dropped" errors on my enp0s25 box. But that's out of 222 million packets, so the rate is quite low, about 1 in 90,000 going missing.

Pinging anything Google related won't be definitive. Due to a problem with Cox infrastructure a few hops from my router, I've been getting 40% packet loss to Google stuff off and on for the past few weeks. This loss also impacts my VPN connection to work... and NOTHING ELSE. I blamed my internal network or WAN connection for a bit before a traceroute showed nothing bad there. Definitely ensure the problem is internal to your network. Move stuff around temporarily and follow the other advice here if you can. Do a traceroute to figure out where the problem is while streaming.

Welp, Spectrum said I could go to a store, explain my sitch and ask for a new modem with a different chipset. I hope that does the trick.

TheGameguru wrote:

That IP you are pinging is a Google DNS server not your router. If that is your router IP then your configuration is super weird.

Oh that was another set of pings I did. All the pings to the router were less than 1 millisecond.

It's a pain, but after troubleshooting connections in old school buildings with ancient wire I have learned that while it is easy to blame sub-par wiring or old physical connections it's a good idea to test cutting all that out first before spending money on infrastructure. Move as close as you can to your router and plug in directly with a known good cable. If it runs flawlessly, upgrade your longer runs. Same issues? Talk to your provider, which is where you seem to have landed.

Good luck!

So you guys, I got a new modem from Spectrum (Docsis3.1, Broadcom chipset and all that) plus a new ethernet cable coupler and a switch to put near the PC, so now I've a switch at both ends of this long connection. I used the "ping.exe -t 8.8.8.8 | Foreach {"{0} - {1}" -f (Get-Date),$_}" command in Powershell, and the highest ping I got after letting it go for 20 minutes was 22ms, the lowest was 9. I'd say that's progress! Yay!

It makes sense that it would be the modem or the ISP, as Ethernet is extremely robust. You can have problems with it, of course, but the chance of the problem being there is a lot lower than most other places, in my experience.

Assuming your problem goes away, then it was almost certainly the modem. The switch you added certainly won't hurt anything, and will put you back in spec, but Ethernet's pretty forgiving, and the old setup probably wasn't a problem. IIRC, it's designed to go 500 meters total, with 3 powered hops and 2 unpowered ones. The length is the most important part: you were way under 500 meters, so you could get away with a fair bit.

Yeah the new modem has definitely helped. I've gone from at least a dozen unstable events during a four hour stream, according to Twitch Inspector, to maybe 3 or 4. Sometimes even none!

I don't know what they consider an 'unstable event', but if you're seeing actual issues in your streaming, you might want to contact your ISP again and ask them for suggestions as to what's wrong.