Where to start when LAN ping losses?

[NewDisplayName]

Hi,

ive found out that i get ping losses to local ip addresses, ... i cant seem to find out if its only one switch or one router. I tried from different locations.

 

Thing i found out:

-I do loose pings to internet, which is fine, because i use LTE...

- But why then to local ip addresses, that makes no sense to me.

I dont use VLAN.

 

network is:

 

DSL+LTE 1 (Hybrid) -> v

DSL+LTE 2 (Hybrid) -> Draytek bonding router -> Google Wifi -> WLAN

                                                                                              -> Switch 

 

I thought its the GWIFI and sometimes restarting this sh1t is helping, but not always... !?!?!?

 

But i also get ping losses to switch and sometimes even to router.

But not always. I only notice it when i play LoL and the nice thing is, it seem to only happen when i engage...?!

 

Anyone a idea where i can start to look?

Edited May 9, 2018 by nuhll

[John_M]

Check cables, connections and MTUs.

[NewDisplayName]

I cant check all cables, i guess i have atleast 300m cat7 cables... there are cables in the roof, in the garden...

 

I need to find a way to find out where exactly the problem is, or atleast, which part of the network.

 

Could it have anything to do with internet loss? (like could internet loss create lan losses? other way around is clear)

 

MTUs cant be the problems, else i would have gotten other errors execpt not reachable. And i never changed any MTU.

 

I guess the next thing i try is when ping losses happen, restart unraid to see if it makes any difference. (so i could say, something on urnaid is causing this)

Edited May 9, 2018 by nuhll

[tr0910]

I cant check all cables, i guess i have atleast 300m cat7 cables... 
 
I need to find a way to find out where exactly the problem is, or atleast, which part of the network.
 
Could it have anything to do with internet loss? (like could internet loss create lan losses? other way around is clear)
 
MTUs cant be the problems, else i would have gotten other errors execpt not reachable. And i never changed any MTU.
 
I guess the next thing i try is when ping losses happen, restart unraid to see if it makes any difference. (so i could say, something on urnaid is causing this)

When I had 300m of cat 5 on one network I had issues too. Lightning would fry it every time. You're going to have to do some cable testing. Are any runs excessively long? Simplify and test is all I can suggest.

Sent from my Nexus 6 using Tapatalk

[Frank1940]

You might need to add switches to breakup your longer runs so as to boast the signal level.  (That is the basic reason for the max length require for the various LAN cable categories.)  Back in the ancient of days (I am talking almost forty years ago), you could buy repeater/amplifiers.  But I could not find any currently for sale.  (Probably because 5 port GB switches are now about $15 a pop!) 

 

For outside, I would recommend WIFI.   Some folks have actually replaced the stick antennas found on most WIFI gear with high gain direction ones to link detached buildings together.  

[NewDisplayName]

Im very happy with wlan coverage, best i had so far, but the gwifi overall is sh1t for ppl who not only just have on enetwork (multiple subnets).

 

Anyway.

 

The thing which bugs me is:

 

I have LAN ping loss from unraid to draytek router (which is in same subnet and like 2m cable) and also on all other ends of the different cables.

 

So it might be "just" the draytek is dead, but how to proof? It doesnt offer any logs, but everythign seems normal. Gets restartet once a week automatic.

 

What ive read is it could happen if you have "network loops"  - what ever that is.

 

I have a rule to allow gwifi subnet to see my "main" network. (to access from 192.168.0 to 192.168.86)

 

COULD that be the problem? In gwifi you cant setup literally anything.

 

I understand it that way that everything what want to go to 192.168.86 will get redirected to the google wifi address in the subnet from the router.

 

And yes, i have atleast single 50-60m connections. But like i said, problems are everywhere. Its not like crazy ping loss tho. But in games u notice when you cant move 10sec... 

 

Thats today, not a single packet loss from unraid to draytek router....

 

Ping-Statistik für 192.168.0.1:
    Pakete: Gesendet = 12928, Empfangen = 12928, Verloren = 0
    (0% Verlust),
Ca. Zeitangaben in Millisek.:
    Minimum = 0ms, Maximum = 14ms, Mittelwert = 0ms

 

Edited May 9, 2018 by nuhll

[John_M]

A 2m long cable must surely be one of the easier ones to test/replace. Since you have such a lot of cable, why not invest in a cable tester? There are cheap ones that just test basic continuity and check that pairs aren't split and there are expensive ones that send high frequency signals that are representative of those normally present on the cable. Even a cheap one has its uses though. It might not fail all bad cables but any it does fail are definitely bad.

[NewDisplayName]

Yeah, ofc, i could change that cable (but they are fresh, i cant believe they broke, but will do that)

 

I think now, maybe i should switch gwifi direct after draytek router and then switch, so i dont have a need for that rule and all devices are on the same subnet.

 

(will be much work)

[pwm]

3 hours ago, Frank1940 said:

You might need to add switches to breakup your longer runs so as to boast the signal level.

 

If having really long segments, good cable quality is important. It isn't just a question of signal level but how well defined impedance, capacitance etc - with long cable runs there is more issues with reflections in the cable. So long outdoor segments are best handled by running at 100Mbit/s instead of gbit/s.

 

5 hours ago, nuhll said:

I need to find a way to find out where exactly the problem is, or atleast, which part of the network.

Start by scanning every single network card you have - check all the individual counters displayed when you do "ifconfig". And if you have smart or semi-smart switches, also check the error counters for the individual switch ports.

 

The above should give you some indication what link segments that have issues.

 

Another thing - if you have very long cable segments, I guess you have bought cable and crimped the connectors yourself? Double-check that the twisted pairs are correct on all cables - using the individual wires in the wrong order can still work but result in much more transfer errors because you don't get the expected common-mode noise rejection intended by the individual twisted pairs in the cable.

[NewDisplayName]

Sorry. i didnt read that "add switches"... yeah, the longest cables without switches/routers are 30m. And this is completly new and the problem was also there before. All others are <15m.

Edited May 9, 2018 by nuhll

[John_M]

5 hours ago, nuhll said:

i have atleast 300m cat7 cables

 

Are you really using Cat 7 cable? What are you terminating it with? GG45 needs a specialist tool and regular 8P8C "RJ45" connectors are a poor match for the cable.

[NewDisplayName]

Yeah, its normal RJ45. It says cat7, imt not sure if thats really correct.

[John_M]

Category 7 has foil screens on the individual pairs and an overall braid screen. Did you put the ends on yourself? It's a real pain to do because everything is too thick for the connector. There's a YouTube video of a guy showing how to do it and he makes a complete mess of it. I suspect that's where your problems lie. It's a waste, really, because the overall spec of the cable is limited by the minimum spec of the parts, like a chain only being as strong as its weakest link.

[NewDisplayName]

Just a quick question.

 

Could a faulty cable fugg the whole network up?

 

I thought like

 

1 <- cable - > 2 <- cable -> 3

 

if cable between 1 and 2 is faulty the route from 2 to 3 should be okay,am i correct?

 

I didnt do anything to the cables i just bought them on amazon. (and plugged them in )

 

https://www.amazon.de/gp/product/B01KV7ZQ1E/ref=oh_aui_search_detailpage?ie=UTF8&psc=1

Edited May 11, 2018 by nuhll

[pwm]

With switches the problem will be just the single link with the bad cable (and obviously result in issues reaching equipment behind that link).

If you happen to have a  hub somewhere, then all data (including bad data) will be forwarded to all the other cables on the hub.

 

If you have a complicated network, you really should have managed switches (at least with web management interface) so you can keep track of transfer errors on every port of the switches.

[NewDisplayName]

Yeah, at some point i will replace everything with ubnt stuff. (looks really nice, just like unraid)

 

I dont have any HUB in my network since.. 20 years?

 

Good, so my thoughts are correct.

 

So it can only be triple NAT problem, or the draytek is just broken (or cant keep up, what ever), or Google wifi (which i also want to replace with ubnt stuff)

BTW, someone need 3 google wifi?  (its the US version, but supports all languages)

Edited May 11, 2018 by nuhll