VDSL Performance & Tuning

Forums › New Zealand Broadband › VDSL Performance & Tuning

1 | ... | 80 | 81 | 82 | 83 | 84 | 85 | 86 | 87 | 88 | 89 | 90 | ... | 285

393 posts

Ultimate Geek

#824475 23-May-2013 23:00

JamesL: Definitely won't be 2db as I had 9db on ADSL2+, at a guess I thought I'd be on ~12db as I'm about 500m from the cabinet

I'd say that it's misreporting

That would be my guess. In the breakdown by band it has downstream 1 as 2.3db and then downstream 2 as 28db. I don't think there would be that much difference. The Fritz doesn't give attenuation by band but it does have the SNR by frequency and from that I'd estimate that my downstream 2 band is only about 10db less than the downstream 1 band. Other than that the connection is looking good though.

stevehodge

393 posts

Ultimate Geek

#824476 23-May-2013 23:03

hio77:
the thing im weary of, is if the 7340 is different to the 7390..

I suspect not. At least the xml data you posted before matched what I was getting from my 7340. They release firmware for both devices at the about the same time so probably the majority of the code is common to both.

What are you developing in, btw?

hio77

12999 posts

Uber Geek

ID Verified

Trusted

Lizard Networks

#824478 23-May-2013 23:13

stevehodge:
hio77:
the thing im weary of, is if the 7340 is different to the 7390..

I suspect not. At least the xml data you posted before matched what I was getting from my 7340. They release firmware for both devices at the about the same time so probably the majority of the code is common to both.

What are you developing in, btw?

i suspect not too, but there could be very minor differences.. newest version for the 7390 has the min/max lines.. but it would seem from comments the newest 7340 doesnt.. that is ui based, but it is also extra data that could be hooked and made use of ;)

im doing it all in C#, .Net 4.0 Client currently.

#include <std_disclaimer>

Any comments made are personal opinion and do not reflect directly on the position my current or past employers may have.

eXDee

4032 posts

Uber Geek

Trusted

#824486 23-May-2013 23:34

If you need a tester on the latest firmware version, i can assist.
I have VS installed too, if you wanted any debugging if anything went wrong.

hio77

12999 posts

Uber Geek

ID Verified

Trusted

Lizard Networks

#824488 23-May-2013 23:44

eXDee: If you need a tester on the latest firmware version, i can assist.
I have VS installed too, if you wanted any debugging if anything went wrong.

im hoping debugging won't be required, pretty sure i have caught all the possible errors.. main reason why im going to release the first versions with a debug toggle (essentially dumping everything before processing)

that being said, i fully expect something to break..

#include <std_disclaimer>

Any comments made are personal opinion and do not reflect directly on the position my current or past employers may have.

hio77

12999 posts

Uber Geek

ID Verified

Trusted

Lizard Networks

#824508 24-May-2013 04:01

alright, whos first to break it!

http://hio77.com/fritzlog/download/1.0.0.0/fritzboxlog.zip

on first run, it will guide through a set few configurations, this will create a xml configuration file, can simply remove it if you need to configure again..

i have also included a no window build, this is going to be rebuilt as a system service rather than a windowless console application, as im not really happy with the way its implemented... feels a tad hackish.

csv exporting isn't included currently, purely a text dump for now.

this works under the assumption you use password protection on your fritz..

while on that topic, in my browsing i came across a fritzbox in sweden which had a open domain assigned to it, along with no password protection, for interest of graphs..


http://i230.photobucket.com/albums/ee313/hio77/nlfproductionsfritzdsl_zps0b4e3e32.png
http://i230.photobucket.com/albums/ee313/hio77/nlfproductionsfritzspec_zpse42cb9e8.png
http://i230.photobucket.com/albums/ee313/hio77/nlfproductionsfritzstat_zps9be991cb.png

posting just links to avoid anyone scrolling down and thinking its my line, few stats there that would raise an question or two..

if anyones interested in the code itself, ill put it on a public github repository..

4:11am, bounced back to DLM-2 it would seem :/ ah well, 2 more weeks..

#include <std_disclaimer>

Any comments made are personal opinion and do not reflect directly on the position my current or past employers may have.

sidefx

3711 posts

Uber Geek

Trusted

#824528 24-May-2013 08:06

Hmm well, 10 CRCPM definitely looks like the upper limit for DLM1 - I got bounced to DLM2 at 4:10 this mornng

"I was born not knowing and have had only a little time to change that here and there." | Octopus Energy | Sharesies
- Richard Feynman

Equinox IT

Cloud spending continues to surge globally, but most organisations haven’t made the changes necessary to maximise the value and cost-efficiency benefits of their cloud investments. Download the whitepaper From Overspend to Advantage now.

JamesL

956 posts

Ultimate Geek
Inactive user

#824548 24-May-2013 09:04

Am I the only one that thinks that DLM is a little too sensitive?

Is 10 CRCPM actually noticeable to internet activities or does it cause overhead elsewhere?

stevehodge

393 posts

Ultimate Geek

#824650 24-May-2013 12:01

JamesL: Am I the only one that thinks that DLM is a little too sensitive?

Is 10 CRCPM actually noticeable to internet activities or does it cause overhead elsewhere?

No it does seem overly sensitive. I've had quite high errors at times and I've never had a disconnect or or issues as far as I can tell.

sidefx

3711 posts

Uber Geek

Trusted

#824655 24-May-2013 12:06

Yeah, I had a SNR tweak a bit high for a while resulting in ~30 CRCPM and didn't notice any issues or disconnections. It was also spiking to ~50CRCPM yesterday (during 1 hour period) according to the stats graphs and no issues surfing the net, running speedtests or disconnections.

Having said that I guess we still don't know 100% the limits on DLM or how it counts resyncs. It may just be coincidence that a number of people have been dropped when they get around 10 CRCPM.

"I was born not knowing and have had only a little time to change that here and there." | Octopus Energy | Sharesies
- Richard Feynman

stevehodge

393 posts

Ultimate Geek

#824658 24-May-2013 12:10

hio77: alright, whos first to break it!

Looks good here. You've got latency misspelt but everything else seems to be working. What is the DsINP value? Is it different from INPRx?

if anyones interested in the code itself, ill put it on a public github repository..

I'm interested. I might try to get it working for Linux via Mono as I have a Linux server that's on all the time.

eXDee

4032 posts

Uber Geek

Trusted

#824664 24-May-2013 12:24

Up we go! DLM-2 from DLM-4 by the look, my INP is now 0/0.8

edit: I win

First run and its gone. Out for a duck

hio77

12999 posts

Uber Geek

ID Verified

Trusted

Lizard Networks

#824694 24-May-2013 13:00

JamesL: Am I the only one that thinks that DLM is a little too sensitive?

Is 10 CRCPM actually noticeable to internet activities or does it cause overhead elsewhere?

its hellishly sensitive..

10CRCPM, your talking 10 packets.. so 1492/1500 bytes per packet.. no way is it actually noticeable..

stevehodge:
hio77: alright, whos first to break it!

Looks good here. You've got latency misspelt but everything else seems to be working. What is the DsINP value? Is it different from INPRx?

its the tweak value, im unsure of what the value completely converts to right now. trust me to misspell something in there.. double checked just about everything.. simple refactor and thats good ;)

INPRx/Tx refer to your INP values that are on the dsl status page.

stevehodge:

if anyones interested in the code itself, ill put it on a public github repository..

I'm interested. I might try to get it working for Linux via Mono as I have a Linux server that's on all the time.

thinking about building it in mono myself to run in a cron on my file server..

eXDee: Up we go! DLM-2 from DLM-4 by the look, my INP is now 0/0.8

edit: I win
First run and its gone. Out for a duck

wooo! error!... one thing that alarms me, your INP is 0.8, DLM calculations wont do too good on that..

does your fritz use password authentication? seems your erroring right on obtaining the challenge which is even before the sid stage!.. which is odd.. i got the baseline from AVM themselves so that should be correct..

i was catching WebExceptions expecting that to be the only plausible error at that stage.. mind dumping http://fritz.box/cgi-bin/webcm?getpage=../html/login_sid.xml something seems to be iffy there..

#include <std_disclaimer>

Any comments made are personal opinion and do not reflect directly on the position my current or past employers may have.

eXDee

4032 posts

Uber Geek

Trusted

#824704 24-May-2013 13:14

That URL does nothing except redirect to the login page. I reset my password to "admin" and the error remains. I'm on the latest firmware.

hio77

12999 posts

Uber Geek

ID Verified

Trusted

Lizard Networks

#824715 24-May-2013 13:22

eXDee: That URL does nothing except redirect to the login page. I reset my password to "admin" and the error remains. I'm on the latest firmware.

alright, we have found a problem.. 3790 too wasnt it? guess ill have to upgrade mine.. see what damage it causes!

if you use chrome, browse to your dsl stats page (pretty much any except for the one with stats over time graphs..) open the developer tool (ctrl shift i) and click the network tab, it should pop up with something like this

firefox has much the same process, i personally prefer chromes debugging tools.

#include <std_disclaimer>

Any comments made are personal opinion and do not reflect directly on the position my current or past employers may have.

1 | ... | 80 | 81 | 82 | 83 | 84 | 85 | 86 | 87 | 88 | 89 | 90 | ... | 285