Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Dray]

As I understand it, there's a bit more to DSLStats and MyDslWebStats than an expect script

[roseway]

Since that DSLstats program requires a GUI, I was hoping just to write a script to retrieve the stats every hour.

I wrote a script that does that and puts them here: http://archive.webconverger.com/dslstats.zip

I was hoping to try upload them to https://www.mydslwebstats.co.uk/ but the site is down for me.

I've been asked before if I could produce a version of DSLstats which runs without a GUI to upload data to MDWS, but I'm afraid it's not going to be possible. Your script gets a subset of the data from the modem, but it takes a great deal of processing to get it into a form which can be uploaded to MDWS, and only two programs are authorised to upload this data. I'm sorry to be so negative, but I'm afraid that following this route is a non-starter.

A simpler low-cost solution is to run DSLstats on a Raspberry Pi. After it's been set up, the RPi will happily run without a monitor connected, even though it's still running the GUI.

[Ronski]

I was hoping to try upload them to https://www.mydslwebstats.co.uk/ but the site is down for me.

You won't be able to directly upload, the method is quite locked down and only Tony and the two authors of HG612 Stats and DSLStats are privy to the upload specifics, and that won't be changing.

[kitz]

Been racking my brains to think why your upstream blocks are only 30 bytes.    This isnt anything to do with the physical condition of the line, but more to do with how DMT works,  how your actual sync speed is calculated and therefore how much useful data can be passed over the connection.
 
If all is set up correctly then with ADSL1 each bin loaded should give you 32kbps of speed.   On the Interleaved channel using s=1/2 mode this in theory should double to 64kbps of sync per bit loaded in the bin. 

I'll be honest and say its not an issue I've ever come across before.  Framing params are negotiated between the modem and DSLAM and its very in-depth technical stuff that takes a degree in the subject to understand.  There is little info about this in the public domain and to be quite frank some of the equations I have seen make my eyes water.   Wombat and I have been attempting to find the relationship values for VDSL framing params for well over a year now.

I know there are different framing modes that can be set and these have an effect on how much speed you get per bit loaded in the DMT bin.

---------

Can you do me a favour and check the settings in your modem to just rule a couple of things out.   I dont think MTU is going to be the problem as thats more TCP/IP, but there are plenty of cases were a wrongly set MTU can do very strange things to a line that does affect throughput speed.
Its just for elimination more than anything else.

Follow the instructions on this page to get to your DSL settings:
http://www.kitz.co.uk/routers/zyxel_VMG8324-B10A_vdsl_setup.htm

Be very careful Do not change anything.  Do not click apply.  I just want to know how its configured.  Specifically

1) The MTU value
2) The type, mode and encapsulation [see below for an example of mine]

Encapsulation should be PPPoE for the BT Openreach network, but I know from a discussion Ive had with TP-link that PPPoE causes some issues on the TT network and IPoE is recommended.    Bearing in mind that you are not directly using TT, but in view of the fact how your phone no vanished from the BT database which normally indicates LLU provision, and we know that AAISP does sometimes use TT for backhaul.

[hendry]

Hope you found the Zip that updates hourly at http://archive.webconverger.com/dslstats.zip interesting ... bit disappointing about the DSL graphing site. Really sad that they couldn't publish an API and such. If they did want to join the cool kids they would use something like Swagger to do it: http://swagger.io/

Photograph of the NTE5 since yesterday afternoon:
http://s.natalian.org/2016-08-05/test-socket-direct.jpg

Modem settings you asked for:
http://s.natalian.org/2016-08-05/1470404632_2558x1404.png
http://s.natalian.org/2016-08-05/1470404608_2558x1404.png

So I'm not sure what to do now. The line's upload channel is still atrocious.

Thank you for looking into this again,

[burakkucat]

I would like to examine some of the raw data that should be available from the ZyXEL modem/router. When convenient, please --

• establish a ssh connection to a local computer connected to the ZyXEL device
• assuming it is a Unix or Linux kernel based system local to the ZyXEL device, establish a logging screen session (screen -L -U)
• establish a telnet connection to the ZyXEL device, log in and invoke a BusyBox shell
• from the BusyBox shell prompt enter the following ten commands --
  
   * adslcmd info --stats
   * adslcmd info --pbParams
   * adslcmd info --Bits
   * adslcmd info --SNR
   * adslcmd info --QLN
   * adslcmd info --Hlog
   * adslcmd info --show
   * adslcmd info --vendor
   * adslcmd info --vectoring
   * adslcmd info --help
  
  
• terminate the BusyBox shell
• terminate the telnet session
• terminate the logging screen session (<Ctrl-A>K)
• compress the log file (perhaps ZIP ?)
• make the compressed file available for me to download

With that raw data I will then generate four graphs: quiet line noise (QLN), transfer function (Hlog), bit loading per sub-carrier and signal to noise ration per sub-carrier. Sight of those four plots may then give us further insight into the circuit's behaviour.

[roseway]

If they did want to join the cool kids they would use something like Swagger to do it: http://swagger.io/

I doubt if Tony wants to join the cool kids. I certainly don't.

[Dray]

I can't see how swagger would add value

[WWWombat]

Good spot.  Why is the block only 30 bytes? 

The $64,000 question...

Supposedly N= K+R with K being the number of bytes in DMT Frame.
Note that 'B' (no of bytes in the data frame) is very low at only 13 bytes... so 13 bytes of what could be useful data* and yet 16 bytes of redundancy overheads?  Is that right

I think 'K' might apply more to ADSL than VDSL framing.

In the cases we are looking at, we have a special case of M=1, so the RS block holds 1 MUX frame.

We know the RS block is 30 bytes, and the FEC overhead is 16 bytes, leaving 14 bytes for the MUX data frame. Yet, whenever we have M=1, the value B (user bytes in MUX data frame) is one less - 13 bytes in this case, rather than 14. My suspicion is that, IIRC, a MUX data frame is the frame that CRC applies to, so the extra byte could hold a CRC checksum. It is only a theory though, totally unconfirmed.

13 bytes in 30 is not good. 43%.

[hendry]

• establish a telnet connection to the ZyXEL device, log in and invoke a BusyBox shell

Maybe my brain is addled from the morning run, but I couldn't figure out how invoke busybox.

http://s.natalian.org/2016-08-06/burakkucat.txt

[hendry]

I can't see how swagger would add value

A service with a well defined API which other clients can submit to, would make it a better service to the community.

The http://editor.swagger.io/ project provides tooling to help define, document and validate a Restful API.

Just a suggestion.

If you guys tell me what values I should plot over what intervals, I could maybe flex my GNUplot muscles, which are pretty weak.

[Dray]

All the necessary information is made available by DSLstats and MyDslWebStats. Just install DSLstats on a RasbPi and upload to MyDslWebstats on a 24*7*365 basis.

[kitz]

@wombat.   Thanks

Ive only seen 'K' on ADSL lines so you are probably correct.  The does seem to be some relationship with 'B' instead - hence me questioning just 13 bytes of what could be useful data.

I'd been having a google and I only found one other instance on the web of a VDSL line having a very small block size.  Most were either 255/240 although admittedly I didn't spend too long searching.
Purely by co-incidence other one I found, actually belonged to TJ (the guy who started hacking the Eircom modems to install standard Zyxel f/w on them).

His stats are here.  Note how he's actually comparing stats from a HG612 -v- (hacked Eircom)VMG8324.   Both set of stats show the same framing params regardless of modem type.  iirc his ISP is (or at least was) Zen but that was a couple of years ago.   Also note that his is a long line..  but those stats also show G.INP so possibly a Huawei cab. 

[kitz]

Maybe my brain is addled from the morning run, but I couldn't figure out how invoke busybox.

http://s.natalian.org/2016-08-06/burakkucat.txt

I think the relevant commands can be accessed 'normally' via telnet without having to specifically invoke the busybox shell.  Zyxel has been nice and haven't b@stardised any of the BCM commands.
FYI the access to the busybox shell on the VMG8324  is > echo && bash.  Unfortunately the busybox shell soon times out no matter what I've tried to set and it can be frustrating trying to get info out of there.

The telnet commands that should give you the equivalent info which b*cat asked for are:

Code: [Select]

adsl info --stats
adsl info --pbParams
adsl info --Bits
adsl info --SNR
adsl info --QLN
adsl info --Hlog
adsl info --show
adsl info --vendor
adsl info --vectoring
adsl info --help


[burakkucat]

Maybe my brain is addled from the morning run, but I couldn't figure out how invoke busybox.

http://s.natalian.org/2016-08-06/burakkucat.txt

And my thought processes were imprecise due to me being unfamiliar with the ZyXEL device in question. However Kitz has defined the commands that should be used from the telnet prompt (>) --

The telnet commands that should give you the equivalent info which b*cat asked for are:

Code: [Select]

adsl info --stats
adsl info --pbParams
adsl info --Bits
adsl info --SNR
adsl info --QLN
adsl info --Hlog
adsl info --show
adsl info --vendor
adsl info --vectoring
adsl info --help


Looking back at your earlier posts, I see that you have previously shown a subset of the requested information --

Wow, that's again for the thorough post. I'm impressed.

Here is the output of the telnet command: http://s.natalian.org/2016-08-03/adslstats.txt

<snip>

I guess I should have looked at the format of the command you had used to harvest that one set of results. 

By a week from today (Saturday 6th August), there should be a ZyXEL VMG1312-B10D installed in The Cattery for usage and experimental purposes . . .