Print

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

[WWWombat]

Thanks.

Those stats tell me some interesting things, but the most important thing to take from them is that you seem to suffer from some big, but intermittent, problems. The route you are taking with the Pi and MDWS is the best thing for now.

First is to note that G.INP is active on both upstream and downstream, which implies DLM has certainly been involved. And both directions show settings for both INP and INPrein in bearer 0; the latter is relatively rare.

At the bottom of the stats, the behaviour in the bottom 5 sub-groupings (from "Latest 15 minutes" through to "Since Link Time") all show a line that is working relatively benignly. Undoubtedly retransmission is working away, and doing a good job - not much makes it through into the serious statistics (the CRC, ES and SES values).

However, the 1st sub-group ("Total Time") show a line in turmoil. Unavailable times of hundreds of seconds is not good (tens of resyncs), and multiple LOS, LOF and LOM - all indications that *sometimes* the line goes AWOL. It isn't good to see these.

The retransmission counters are high in this snapshot, but they can get that way - it becomes important to see how they change over time (through graphs) rather than a snapshot. We'll leave that until MDWS is running.

[MaximusPrime]

@WWWombat,

cheers for the info, my line resynced last night, up to 7Mb now, I'll look to get my MDWS set up in the next couple of weeks, cheers

[MaximusPrime]

Having a lot of errors on my line today, back down to 4Mb,

Any ideas?

[kitzuser87430]

If there are thunderstorms in the general area (100 miles) ES, FEC and CRC's most probably will increase.

[MaximusPrime]

@kitzuser8743,

we had mild rain today, the problems started, we had thunder after but I didn't see any lighting

If this was the case my line is even more prone to the weather conditions than I previously thought

[MaximusPrime]

@kitzuser8743,

I don't think it was the weather that caused todays problems, the weather cleared up but the errors have continued

[WWWombat]

Its a tricky process, correlating error rates against external factors.

One prospect would be to correlate the error rate against changes in SNRM. And against resyncs, and changes in speed.

Another prospect is to correlate the FECs, CRCs and ESs against the various retransmission counters.

Another prospect is to correlate them against behaviour from the same time yesterday, or the day before. Or against Monday a week ago.

And another prospect is to correlate them against alterations in the DLM settings, and the various framing parameters. Especially when resyncs keep happening - when we need to figure out why those happen too.

Right now, we're getting lots of little, partial, snapshots. Its a bit like trying to figure the landscape through a fogged windscreen with a tiny bit cleared by the demister...

[MaximusPrime]

@WWWombat,

I haven't seen my line so bad in a long time. I usually have no CRC's. It'll take me a couple of weeks to research, order & setup my PI.

On another topic, what's your opinion in making my own RJ45 to RJ11 cable like in this vid: https://www.youtube.com/watch?v=trFUjOPPutM&index=1&list=PLexbWs0Wp_H6LvKLZ6n5fyUknvlMag-Su any benefit?

[WWWombat]

I couldn't say, as I've never needed to try anything like that.

However, others on here have gone down that path, best described here:
http://forum.kitz.co.uk/index.php/topic,15910.0.html

[MaximusPrime]

My HLog looks messed up, any observations?

I'll order my Raspberry Pi this week & setup Web DSL Stats next week

[burakkucat]

My HLog looks messed up, any observations?

Hmm . . . I wouldn't describe it as messed up. The staircase effect is just an artefact of the scale of the plot versus the granularity of the available data. What is of concern, is that you only have one upstream band available (US0) and two downstream bands available (DS1 & DS2).

The two snapshot graphs that you have attached show that the circuit is in a bad way. (Which really states the obvious, in the light of this thread's subject line.) Once you have a R-Pi setup and regularly harvesting data for uploading to MDWS then there will be sufficient information for WWWombat, say, to analyse.

[MaximusPrime]

@burakkucat,

I set up a MDWS account today, it's on a temporary Linux box. My username is 'MaxPrime'.

According to stats I only have 1 downstream band 'D1'. That must be even worse? What is a band?

[burakkucat]

xDSL circuits operate by a frequency division duplex process.

With G.992.1, G.992.3 & G.992.5 (i.e. G.DMT, ADSL2 & ADSL2+) there is an US band that starts above (at a higher frequency) the PSTN (analogue telephony). It is from (about) 30 kHz to (about) 138 kHz. Then from (about) 138 kHz upwards to the upper frequency limit (depending upon the mode; the G.992.x specification) is the DS band. So for those three modes, there is just one US and one DS band.

With G.993.2 (i.e. VDSL2) there are multiple US and DS bands, the actual number allocated for use is set by the profile. The current VDSL2 profile in the UK is 17a. That profile defines three US bands (labelled US0, US1 & US2) and three DS bands (labelled DS1, DS2 & DS3). Those six bands are interleaved across the entire (approximate) 17 MHz bandwidth of profile 17a. From low to high frequencies the sequence is US0, DS1, US1, DS2, US2 & DS3.

Looking at your two shapshot graphs, the Hlog plot indicated to me that the circuit had the potential to use US0, DS1 & DS2 and that also appeared to be the case when the Bit loading plot was also considered.

Look closely at the information reported by the xdslcmd info --pbParams command and you will see information about two band plans displayed. The first is the discovery phase and the second is the medley phase. The initial, discovery, phase shows the bands that could be usable . . . as determined by the modem and DSLAM whilst they are performing their preliminary ritual when achieving synchronisation. The final, medley, phase (also known as "showtime") displays the bands that are actually usable. In your case, you only have one US and one DS band available for use . . . so it is effectively operating just "slightly better" than an ADSL2+ circuit. <Ouch!> 

[MaximusPrime]

@burakkucat,

here's my pbParams:

adsl info --pbParams
adsl: ADSL driver and PHY status
Status: Showtime
Last Retrain Reason:    8000
Last initialization procedure status:   0
Max:    Upstream rate = 1126 Kbps, Downstream rate = 8284 Kbps
Bearer: 0, Upstream rate = 800 Kbps, Downstream rate = 7064 Kbps
Bearer: 1, Upstream rate = 0 Kbps, Downstream rate = 0 Kbps
Discovery Phase (Initial) Band Plan
US: (7,32) (871,1205) (1972,2782)
DS: (33,859) (1216,1961) (2793,3970)
Medley Phase (Final) Band Plan
US: (7,32)
DS: (33,707)
                  VDSL Port Details               Upstream                Downstream
Attainable Net Data Rate:            1126 kbps               8284 kbps
Actual Aggregate Tx Power:            10.5 dBm                9.2 dBm
====================================================================================
  VDSL Band Status   U0   U1   U2   U3   U4   D1   D2   D3
  Line Attenuation(dB):   18.4   N/A   N/A   N/A   N/A   36.7   N/A   N/A
Signal Attenuation(dB):   18.4   N/A   N/A   N/A   N/A   58.8   N/A   N/A
        SNR Margin(dB):   13.9   N/A   N/A   N/A   N/A   7.5   N/A   N/A
         TX Power(dBm):   10.5   N/A   N/A   N/A   N/A   9.3   N/A   N/A
 >

Tones 33 to 707 DS should be stable then? But even some of those are missing. Do you know any reasons why they are missing & how I could get more tones?

cheers

[burakkucat]

Looking at what that "pbParams" output shows --

Bearer: 1, Upstream rate = 0 Kbps, Downstream rate = 0 Kbps

That is the good news; G.Inp is active on your circuit. 

Bearer: 0, Upstream rate = 800 Kbps, Downstream rate = 7064 Kbps

That is the bad news; it shows your circuit has the bandwidth of G.992.3 (i.e. ADSL2). 

Medley Phase (Final) Band Plan
US: (7,32)
DS: (33,707)

Confirmed by the "showtime" band plan. 

The above was the easy part.

The difficult part now begins. Once sufficient data has been recorded, analysing it to understand what is causing the circuit to operate so poorly. At the moment, nobody can say.