Intermittent massive ping times

[Lee Nowell]

Hi All,

I have been having sporadic issues with my internet. A couple of times a day, when we are streaming on demand TV, it will pause and the buffering circle appears. After a minute or so it will sort itself out and continue.

I have noticed that when this happens, pinging the Google DNS server (8.8.8.8) I get massive ping times usually between 1,000 and 5,000 ms! Looking at this in more detail over the past few days I have run a couple of tests
1. Continually ping'd 8.8.8.8 over 24 hours. Looking at the log, the issue seems to happen a few times an hour
2. I have run traceroute 8.8.8.8 for a few hours and I have a few instances of the long ping times. I have pasted the traceroute for the 3 occasions below (changed my internal addresses to "MYIP")
First instance
Mon 3 May 11:39:18 BST 2021
traceroute to 8.8.8.8 (8.8.8.8), 64 hops max
1 <MYIP1> 0.172 0.151 0.159
2 <MYIP2> 0.615 0.692 0.686
3 * * *
4 213.121.98.129 2362.102 * *
5 * 213.121.98.128 138.378 *
6 87.237.20.138 3360.186 2716.418 2742.804
7 87.237.21.174 2935.524 * 110.855
8 * * *
9 8.8.8.8 2998.063 * 582.409

Second instance
Mon 3 May 12:09:57 BST 2021
traceroute to 8.8.8.8 (8.8.8.8), 64 hops max
1 <MYIP1> 0.212 0.17 0.149
2 <MYIP2> 0.689 0.683 0.682
3 * * *
4 * * *
5 * 213.121.98.128 560.209 2635.551
6 * 87.237.20.138 7.975 *
7 87.237.21.174 2589.703 2794.373 2955.29
8 * * *
9 * 8.8.8.8 482.195 *

Third instance
Mon 3 May 13:49:45 BST 2021
traceroute to 8.8.8.8 (8.8.8.8), 64 hops max
1 <MYIP1> 0.188 0.146 0.139
2 <MYIP2> 0.71 0.741 0.479
3 * * *
4 * * 213.121.98.129 376.329
5 * 213.121.98.128 272.967 2467.027
6 87.237.20.138 1047.391 * 286.63
7 87.237.21.174 2626.708 1960.504 1646.555
8 * * *
9 8.8.8.8 8.795 8.91 9.195

As you can see they vary a little and also the impact of the overall ping differs (e.g. the third one has a good end to end ping time but the intermediate ones are poor)

Anyone have any ideas how to trouble shoot this? When I spoke with the ISP, they said that they can not work on ping times only run the line tests. I am trying to get to a point where I can give them something concrete to go on.

Thanks in advance

Lee.

[Graham J]

Lee Nowell wrote:
> Hi All,
>
> I have been having sporadic issues with my internet. A couple of times a day, when we are streaming on demand TV, it will pause and the buffering circle appears. After a minute or so it will sort itself out and continue.

[snip]

Who is your ISP?

What router do you have?

Does it show you the traffic generated by each user? Is there any
evidence of something on your LAN uploading significant volumes of data
on the occasions when you see the buffering?

Do you have ADSL, FTTC, or FTTP?

If ADSL or FTTC does the router show any indication of retries or sync
failures on the occasions when you see the buffering?

Can you ping your ISP's gateway IP in simultaneously with the pinging
the Google DNS? There may be a graphing tool that will do this so you
can save the results in an easily viewable form. If the gateway
responds reliably but Google does not then there may be a problem at
your ISP.



--
Graham J

[Lee Nowell]

Hi Graham,

I have tried to answer your questions inline but please let me know if it doesn't come through correctly and I will redo it - I have had problems in the past.

On Monday, 3 May 2021 at 17:42:57 UTC+1, Graham J wrote:
> Lee Nowell wrote:
> > Hi All,
> >
> > I have been having sporadic issues with my internet. A couple of times a day, when we are streaming on demand TV, it will pause and the buffering circle appears. After a minute or so it will sort itself out and continue.
> [snip]
>
> Who is your ISP?

EE

>
> What router do you have?
>

EE Brightbox

> Does it show you the traffic generated by each user? Is there any
> evidence of something on your LAN uploading significant volumes of data
> on the occasions when you see the buffering?

I don't believe the router does but I run ClearOS as the internal gateway which all internet traffic goes through before hitting the router which is on a different subnet to everything else. ClearOS may well have some logs - will take a look but I don't believe so as the only thing that uploads anything significant is the backup and that is all up to date.

>
> Do you have ADSL, FTTC, or FTTP?

ADSL (router shows it as G.992.5 (ADSL2+))

>
> If ADSL or FTTC does the router show any indication of retries or sync
> failures on the occasions when you see the buffering?

Not that I can see - the system log just has time sync enties, no sync, disconnects etc.

>
> Can you ping your ISP's gateway IP in simultaneously with the pinging
> the Google DNS? There may be a graphing tool that will do this so you
> can save the results in an easily viewable form. If the gateway
> responds reliably but Google does not then there may be a problem at
> your ISP.

I am running pingplotter so will try and find out EE's gateway address and add that to the list.

Thanks again for your help

Lee.
>
>
>
> --
> Graham J

[Lee Nowell]

Quick update. Looking at the router, in the internet section, it gives me the IP address of the gateway - oddly, when I ping this I get no response! The primary and secondary DNS server ping fine and seem to have the same issue as pinging 8.8.8.8. I just had ping times go to 1.1 seconds in pingplotter and both DNS servers and 8.8.8.8 all went to more or less the same value.

[Graham J]

Lee Nowell wrote:
> Hi Graham,
>
> I have tried to answer your questions inline but please let me know if it doesn't come through correctly and I will redo it - I have had problems in the past.

Inline answers show OK, thanks

[snip]

>> Who is your ISP?
>
> EE

Oh dear ...

>> What router do you have?
>>
>
> EE Brightbox

OK - pretty basic
[snip]

> I don't believe the router does but I run ClearOS as the internal gateway which all internet traffic goes through before hitting the router which is on a different subnet to everything else. ClearOS may well have some logs - will take a look but I don't believe so as the only thing that uploads anything significant is the backup and that is all up to date.

Are you sure all you traffic gose through ClearOS? Have you disabled
the WiFi service in the Brightbox?
[snip]

More in next reply ...


--
Graham J

[Graham J]

Lee Nowell wrote:

[snip]

>> I am running pingplotter so will try and find out EE's gateway address and add that to the list.

> Quick update. Looking at the router, in the internet section, it gives me the IP address of the gateway - oddly, when I ping this I get no response!

That may be a deliberate choice by EE. However you should see that IP
in a traceroute to something on the internet (e.g. the Google DNS)

> The primary and secondary DNS server

Are these the DNS servers operated by EE?

> ping fine and seem to have the same issue as pinging 8.8.8.8. I just had ping times go to 1.1 seconds in pingplotter and both DNS servers and 8.8.8.8 all went to more or less the same value.

A router that you can get to work with RouterStats - see:

<http://www.vwlowen.co.uk/internet/files.htm>

... could log any changes in SNR margin which might indicate bursts of
noise which cause re-tries or potentially loss of sync.

Can you get an upgrade to FTTC?

Have you thought of changing ISP? Zen Internet would be my
recommendation; alternatively Andrews & Arnold. Both these ISPs have
technical support who would understand the issues and work with you to
resolve problems.


--
Graham J

[Lee Nowell]

Yes all traffic goes through Clear OS - wifi and dhcp etc disabled on the brightbox

[Lee Nowell]

On Monday, 3 May 2021 at 20:24:42 UTC+1, Graham J wrote:
> Lee Nowell wrote:
>
> [snip]
> >> I am running pingplotter so will try and find out EE's gateway address and add that to the list.
> > Quick update. Looking at the router, in the internet section, it gives me the IP address of the gateway - oddly, when I ping this I get no response!
> That may be a deliberate choice by EE. However you should see that IP
> in a traceroute to something on the internet (e.g. the Google DNS)

Oddly no sign of the "internet gateway" IP address in the traceroute. Also, no sign of the EE DNS servers in pingplotter other than when I add them specifically to the ping test

> > The primary and secondary DNS server
> Are these the DNS servers operated by EE?
> > ping fine and seem to have the same issue as pinging 8.8.8.8. I just had ping times go to 1.1 seconds in pingplotter and both DNS servers and 8.8.8.8 all went to more or less the same value.
> A router that you can get to work with RouterStats - see:
>
> <http://www.vwlowen.co.uk/internet/files.htm>
>
> ... could log any changes in SNR margin which might indicate bursts of
> noise which cause re-tries or potentially loss of sync.
>
> Can you get an upgrade to FTTC?

I can although our cabinet is about 1/4 mile from the exchange

>
> Have you thought of changing ISP? Zen Internet would be my
> recommendation; alternatively Andrews & Arnold. Both these ISPs have
> technical support who would understand the issues and work with you to
> resolve problems.

I wanted to see if I could prove to EE there is an issue first and let them try to resolve before jumping ship
>
>
> --
> Graham J

[Martin Brown]

What does the advanced DSL status info show? Sync and error rates?

Various forms of uncorrectable error and hard error seconds can result
in exponential backoff which becomes quite disastrous for streaming.

When mine goes haywire hard error seconds increment in real time but the
"bright" box isn't smart enough to notice. Their provided router is just
marginally faster than my proper router so I have stuck with it.

Transient line faults isn't something it handles well.

It is a dry spring so I get 1M up and 7M down on rural wet string ADSL
2+ (subject to today's rain).

--
Regards,
Martin Brown

[Lee Nowell]

Thanks for your reply Martin. DSL Status page is as follows


DSL Status
This page shows information about your DSL connection. If you are using Fibre/Ethernet Broadband then this page is not applicable.
Status
Configured Current
Line Status -- UP
Link Type -- Fast Path
Operation Mode Automatic G.992.5 (ADSL2+)
Data Rate Information
Upstream 1251 (Kbps)
Downstream 14167 (Kbps)
Defect/Failure Indication
Operation Data Upstream Downstream
Noise Margin 6.2 (dB) 3.1 (dB)
Line Attenuation 15.0 (dB) 30.0 (dB)

Indicator Name Near End Indicator Far End Indicator
Output Power 12.3 (dBm) 0.0 (dBm)
Fast Path FEC Correction 0 0
Interleaved Path FEC Correction -- --
Fast Path CRC Error 14063 155
Interleaved Path CRC Error -- --
Loss of Signal Defect 0 0
Fast Path HEC Error STR 41185 209
Interleaved Path HEC Error -- --
Error Seconds 7882 104
Statistics
Received Data 34359738 (Kbits)
Transmitted Data 11866468 (Kbits)

Thanks

Lee.

[Graham J]

Lee Nowell wrote:

[snip]

>> Can you get an upgrade to FTTC?
> I can although our cabinet is about 1/4 mile from the exchange

For FTTC this important parameter is the distance from you to the
cabinet. From your other reply the line attenuation at 30dB suggests
the exchange is 2km away, so the cabinet being a quarter mile nearer
would give a line of about 1.6km, so the ISP won't offer FTTC at that
distance.

>> Have you thought of changing ISP? Zen Internet would be my
>> recommendation; alternatively Andrews & Arnold. Both these ISPs have
>> technical support who would understand the issues and work with you to
>> resolve problems.
>
> I wanted to see if I could prove to EE there is an issue first and let them try to resolve before jumping ship

You could ring Zen. Ask initially for sales. Explain the nature of the
problem and that you are considering moving. Ask to speak to somebody
technical who can discuss what Zen would do to diagnose and hence
resolve the problem if they were your ISP. This is how I first got to
use Zen, where the existing ISP had not been able to resolve the
problem, not even understand it.


--
Graham J

[Andy Burns]

Lee Nowell wrote:

> Graham J wrote:
>
>> Can you get an upgrade to FTTC?
>
> I can although our cabinet is about 1/4 mile from the exchange

With FTTC, what counts is distance from home to cabinet, exchange to
cabinet is irrelevant, and the fibre probably goes to a bigger exxchange
further away then the copper goes to anyway

I'd say anyone on ADSL, who has the option of VDSL should jump at it,
reliability and speed will increase, price difference is minimal.

[Andy Burns]

Lee Nowell wrote:

> Fast Path CRC Error 14063 155

> Fast Path HEC Error STR 41185 209

> Error Seconds 7882 104

yuck, yuck, yuck.

[Graham J]

Lee Nowell wrote:

[snip]

> Thanks for your reply Martin. DSL Status page is as follows
>

Relevant results:

> Link Type -- Fast Path

> Fast Path FEC Correction 0 0

> Fast Path CRC Error 14063 155

> Fast Path HEC Error STR 41185 209

> Error Seconds 7882 104

The key here would be to see whether these error counts increase rapidly
at the same moment you get the buffering; which would confirm
intermittent noise. I think you can tailor RouterStats to capture these
parameters - but it depends on the router.

Normally I would suggest getting a decent router in order to run the
tests, but it might be easier to change to Zen and use the (probably
free) router they will send you.


--
Graham J

[Lee Nowell]

lol - that doesn't sound good :) Does that suggest there are issues with my ISP / line?

[Lee Nowell]

Ok thanks Graham - will keep an eye on the tests to try and monitor these when I get the issue.

[Lee Nowell]

Quick update. I have had 3 issues of long ping times but pingplotter has them lasting a second or so each. The error numbers have increased but not by much. Now

Fast Path FEC Correction 0 0

Interleaved Path FEC Correction -- --

Fast Path CRC Error 14149 158

Interleaved Path CRC Error -- --
Loss of Signal Defect 0 0

Fast Path HEC Error STR 41448 213

Interleaved Path HEC Error -- --

Error Seconds 7918 107

[NY]

"Lee Nowell" <leen...@yahoo.co.uk> wrote in message
news:4870c6ad-5f10-4348...@googlegroups.com...

What would you diagnose with a VDSL line which varies (due to spontaneous
router reconnection at around 2 AM) between 20 and 30 Mbps. Mostly around 20
but sometimes it rises to 30.

Attenuation remains pretty constant at about 27 dB up and 16 dB down, but
noise margin reduces at times of higher speed in the relevant direction:


Normal figures are about

date/time | sync (up/down: kbps) | noise margin U/D (dB) | line atten
U/D (dB) | Ookla speedtest data rate (U/D: Mbps)

21/05/2020 10:35 | 9319 28678 | 6.0 11.5 | 27.9 17.3 | 8.41
26.68


Extremes have been

25/06/2020 14:30 | 12391 23830 | 3.1 14.3 | 27.8 17.1 | 11.42
22.36(

lower than normal NM upstream with consequent increase in upstream sync and
data-transfer rate)


30/03/2021 16:00 | 9464 34956 | 5.9 8.6 | 27.2 16.6 | 8.64
31.73
02/05/2021 18:40 | 9254 34956 | 6.0 8.9 | 26.9 16.0 | 8.50
32.31

(lower than normal NM downstream with consequent increase in downstream sync
and data-transfer rate)


Router is Plusnet Hub One
(https://www.plus.net/help/assets/images/broadband/routers/hub-one.png)



I'm not certain where my cabinet is, but I know there's one about 1/4 mile
(400 m) away.
using modulation G.993.2 Annex B. This router doesn't give any other
diagnostics such as error rates.

[Roderick Stewart]

On Mon, 3 May 2021 21:53:38 +0100, Andy Burns <use...@andyburns.uk>
wrote:

>I'd say anyone on ADSL, who has the option of VDSL should jump at it,
>reliability and speed will increase, price difference is minimal.

Absolutely, and buy a decent modem/router of your own. Don't just rely
on what your ISP can offer. Many of them can handle both ADSL and VDSL
so you don't even need to wait. You can install it whenever convenient
to you, and if you later opt for VDSL you'll only need to change a few
settings on the day.

Rod.

[Graham J]

NY wrote:

[snip]

> What would you diagnose with a VDSL line which varies (due to spontaneous
> router reconnection at around 2 AM) between 20 and 30 Mbps. Mostly
> around 20 but sometimes it rises to 30.

Noise.

The reconnections around 2 am suggest the Openreach system is trying to
optimise your connection, but never getting any better. A few
reconnections per year at this time of night will be Openreach rebooting
something at their end. But on a good quiet line the connection can
remain up for several thousand hours at a time, even with ADSL.

> Attenuation remains pretty constant at about 27 dB up and 16 dB down, but
> noise margin reduces at times of higher speed in the relevant direction:

Downstream attenuation at 16dB suggests about 1.2 km to cabinet. This
is consistent with about 28 MBits/sec downstream sync.

For downstream, the SNR margin tries to achieve about 6dB, and will show
maximum speed. If there is noise the SNR margin will increase with
consequent drop in sync speed.

Ths does not apply to the upstream direction, because the upstream speed
is usually capped by design. So the router will sometimes show a larger
SNR margin - essentially this information is not useful.

> Normal figures are about
>
> date/time  |  sync (up/down: kbps)  |  noise margin U/D (dB)  |  line atten
> U/D (dB)  |  Ookla speedtest data rate (U/D: Mbps)
>
> 21/05/2020 10:35  |  9319   28678  | 6.0    11.5  |  27.9   17.3  |
> 8.41 26.68

Downstream SNR margin at 11.5 dB suggests the line is noisy.

> Extremes have been
>
> 25/06/2020 14:30  |  12391 23830  |  3.1  14.3  |  27.8 17.1  |  11.42
> 22.36
>

> lower than normal NM upstream with consequent increase in upstream sync and
> data-transfer rate)

The noise may impact only part of the available spectrum so your router
optimises where it can. Here it has increased the upstream speed at the
expense of downstream.

[snip]

> Router is Plusnet Hub One
> (https://www.plus.net/help/assets/images/broadband/routers/hub-one.png)
>
> I'm not certain where my cabinet is, but I know there's one about 1/4
> mile (400 m) away.

As discussed above - a bit over 1 km. Look at:
<https://kitz.co.uk/adsl/cabinet-lookup.htm#fttc_cab>

> using modulation G.993.2 Annex B. This router doesn't give any other
> diagnostics such as error rates.

So get a decent router - Draytek 2760 or similar - to get good diagnostics.


--
Graham J

[Martin Brown]

OK. A couple of things (one I have only just half remembered from a
previous EE support call about my streaming misbehaving on Win7). It
happened after getting my first Win7 machine so a long time ago now.

These are my statistics for the second block:

Fast Path FEC Correction -- --
Interleaved Path FEC Correction 216864 300732
Fast Path CRC Error -- --
Interleaved Path CRC Error 1090 302
Loss of Signal Defect 5 0
Fast Path HEC Error STR -- --
Interleaved Path HEC Error 12630 723
Error Seconds 998 474

Statistics
Received Data 34359738 (Kbits)

Transmitted Data 7437770 (Kbits)

Note that mine is on interleaved error correction - my line is far too
unstable to support fastpath at all. If you don't do action gaming then
asking your provider to force interleaved error correction may help.

The price I pay is that ping increases from 25ms to 40ms (so useless for
gaming) and the result is a network connection that remains stable even
when it is raining hard. My phone line is very exposed. Hard failures
have included trees rubbing the insulation off or snapping it entirely.

The other is that one of Microsoft's services impacts streaming internet
data in a peculiar way if you happen to have a network shared printer of
the wrong sort. I can't recall the exact details but a call to support
describing your intermittent streaming woes might well get results.

I am struck by our coincidental data download figures being identical -
it seems to be clamped at that very odd maximum value! (mine hasn't
changed at all since I noticed this and I am using data at a fair rate).

Do any other routers clamp displayed data usage at weird values?
This odd value is a little *over* 2^25 = 33554432

And it doesn't divide by 8 either factorised as (2.3.7.199.4111)

--
Regards,
Martin Brown

[NY]

"Graham J" <nob...@nowhere.co.uk> wrote in message
news:s6qvc1$aai$1...@dont-email.me...

> As discussed above - a bit over 1 km. Look at:
> <https://kitz.co.uk/adsl/cabinet-lookup.htm#fttc_cab>

I tried that before I posted my original question, but the site doesn't seem
to be working properly. I get a large empty square on the
https://kitz.co.uk/adsl/btwchecker.php page:
https://i.postimg.cc/q7w9MsVX/Screenshot-2021-05-04-Kitz-BTw-FTTC-availability-checker.png
where I'd expect to be asked my phone number and postcode.

https://kitz.co.uk/adsl/broadbandchecker.php linked as "broadband checker"
on that page, prompts for phone number / postcode and gives details about my
exchange, but not the cabinet I'm connected to.

I've tried this in Firefox 89 and Chrome.


You mention line noise. Is it likely to be noise rather than
distortion/selective attenuation? I ask because I've always been suspicious
of the wiring within the house. We have no obvious master socket (none of
the sockets have removable faceplates) and the route of the cable within the
house, from one socket to the next, is a mystery between the GPO lozenge box
on the gable end near where the overhead cable meets the house and the
closest socket.

When I can find a time to unplug the router without disturbing my wife's
working-from-home, I might try the router in the other sockets to see if
there's any significant difference to noise margin and attenuation, though I
won't have an easy way of disconnecting downstream wiring, apart from
pulling wires out of the IDC terminals.

Weird that the router disconnects overnight every few days - I agree it's as
if it knows it might be able to do better, and then finds that it can't.

I have a TPlink TD-W9980 router from our previous house (I can't remember
why we use the Plusnet router rather than the Tplink in this house...) which
might give better diagnostics. I'll give it a try.

30/9 Mbps is not too bad (we pay for "up to 40") but 20/9 is not so good.

[Graham J]

NY wrote:
> "Graham J" <nob...@nowhere.co.uk> wrote in message
> news:s6qvc1$aai$1...@dont-email.me...
>> As discussed above - a bit over 1 km.  Look at:
>> <https://kitz.co.uk/adsl/cabinet-lookup.htm#fttc_cab>

[snip]

Try <https://www.broadbandchecker.btwholesale.com/#/ADSL>

> You mention line noise. Is it likely to be noise rather than
> distortion/selective attenuation? I ask because I've always been
> suspicious of the wiring within the house. We have no obvious master
> socket (none of the sockets have removable faceplates) and the route of
> the cable within the house, from one socket to the next, is a mystery
> between the GPO lozenge box on the gable end near where the overhead
> cable meets the house and the closest socket.

Your description suggests there may be "bridge taps", see:

<https://en.wikipedia.org/wiki/Bridge_tap>

These cause signal refelections which may reduce the available spectrum
for your VDSL signal. But this is likely to be constant, whereas your
performance varies - which suggests intermittent noise.

May well be worth paying Openreach to regularise your wiring with a
master socket at the most convenient location.

[snip]

Do try the TP-Link router ...

--
Graham J

[NY]

"Graham J" <nob...@nowhere.co.uk> wrote in message

news:s6r457$cd3$1...@dont-email.me...

> NY wrote:
>> "Graham J" <nob...@nowhere.co.uk> wrote in message
>> news:s6qvc1$aai$1...@dont-email.me...
>>> As discussed above - a bit over 1 km. Look at:
>>> <https://kitz.co.uk/adsl/cabinet-lookup.htm#fttc_cab>
>
> [snip]
>
> Try <https://www.broadbandchecker.btwholesale.com/#/ADSL>

Grrr. It won't give me a result my although the line is provided by BT, the
phone service is provided by Plusnet.

>> You mention line noise. Is it likely to be noise rather than
>> distortion/selective attenuation? I ask because I've always been
>> suspicious of the wiring within the house. We have no obvious master
>> socket (none of the sockets have removable faceplates) and the route of
>> the cable within the house, from one socket to the next, is a mystery
>> between the GPO lozenge box on the gable end near where the overhead
>> cable meets the house and the closest socket.
>
> Your description suggests there may be "bridge taps", see:
>
> <https://en.wikipedia.org/wiki/Bridge_tap>
>
> These cause signal refelections which may reduce the available spectrum
> for your VDSL signal. But this is likely to be constant, whereas your
> performance varies - which suggests intermittent noise.
>
> May well be worth paying Openreach to regularise your wiring with a master
> socket at the most convenient location.

Yes I'd wondered about that. It's one of those grey areas: is the reduction
in speed bad enough to warrant paying BTOR to sort out the wiring. I'll have
a look in the loft first of all and see if I can see where the wiring goes
after it leaves the GPO box on the outside wall. If there is one cable, the
sockets may be daisy-chained, but if there is more than one, I may have the
dreaded star-connected topology which worked fine for phones (the wiring was
installed long before DSL was ever thought of) but causes reflections for
higher frequencies.

Come to think of it, we have an extension bell which used to work and then
stopped when we had some building work down which involved the electrician
replacing an old, cracked BT socket (not the one used for the router) with a
new one. I wonder if the reduction in speed dates from then. I have a vague
memory that the sync speed used to be a bit faster before that. There's no
harm in keeping things as simple as possible: we don't need either of the
other sockets because the one for the router is also the one for the DECT
base station.

The only complication might be getting access to that GPO socket on the
outside wall, because the only access to it is via a sloping tiled roof
(it's on a short vertical wall between two ridged roofs at different
heights). I remember when the BTOR engineer was tracing our line because of
problems finding a vacant pair from the pole to the cabinet, he said he'd
have liked to monitor at the GPO socket but couldn't get access to it
because his "ticket" (H&S?) didn't cover it. I bet it needs extra H&S skills
to work on a sloping ladder along a tiled roof, compared with working up a
tall vertical ladder from the ground.

But I take your point about reflections causing a constant problem rather
than something that varies - unless the conditions are borderline and
trigger the router and DSLAM at the cabinet to sync in one of two states
(higher sync with lower NM, or lower sync with higher NM) depending on
phases of the moon and whether there's an R in month. ;-)

> [snip]
>
> Do try the TP-Link router ...

Ah. I've just booted that router (without connecting it to the phone line)
and it looks as if it only gives the same info (NM, atten, sync speed) as
the Plusnet router - unless extra info/menus become available once a DSL
carrier is detected. It's a few years now since I used that router, so I
can't remember what diagnostics it gives. I'll try it again this evening
when the interruption won't be noticed.

[grinch]

On 04/05/2021 09:15, Martin Brown wrote:
> On 03/05/2021 21:07, Lee Nowell wrote:
>> On Monday, 3 May 2021 at 21:01:34 UTC+1, Martin Brown wrote:
>>> On 03/05/2021 18:57, Lee Nowell wrote:
>>>> Hi Graham,
>>>>
>>>> I have tried to answer your questions inline but please let me know
>>>> if it doesn't come through correctly and I will redo it - I have had
>>>> problems in the past.
>>>>
>>>> On Monday, 3 May 2021 at 17:42:57 UTC+1, Graham J wrote:
>>>>> Lee Nowell wrote:
>>>>>> Hi All,
>>>>>>
>>>>>> I have been having sporadic issues with my internet. A couple of
>>>>>> times a day, when we are streaming on demand TV, it will pause
>>>>>> and the buffering circle appears. After a minute or so it will
>>>>>> sort itself out and continue.
>>>>> [snip]
>>>>>
>>>>> Who is your ISP?
>>>>
>>>> EE
>>>>>
>>>>> What router do you have?
>>>>>
>>>>

This is a simple network traffic issue.You have stated that your open
reach adsl circuit works some of the time so wasting money on another
router will achieve little.

As openreach provided dsl is layer2 you cant see if there are any
traffic congestion issues from home users point of view.And openreach
will always deny any network issues unless the network is hard down.

Try winmtr to give you a rough idea where is congestion is but loosing
pings does not necessarily prove packet loss as ICMP is seen as
unimportant traffic by most makes of PE router.

Moving to another ISP with less congestion might help and the Zen
provided Fritzboxes are a good router.

I don't have congestion issues with my Zen account but most of my home
network is Cisco.I am on Zen FTTC with a working dual stack.

[Lee Nowell]

On Monday, 3 May 2021 at 21:59:56 UTC+1, Graham J wrote:

This thread seems to have split with a separate issue to my original one so hopefully I am replying to the correct bit of the thread. I spoke with EE earlier and they are sending out a Qube engineer on Thursday. Not sure what one of these is but are there any useful things I could monitor/ track between now and then which would help them troubleshoot the issue? Maybe how the results above change over time - assume I will need to do this manually on the router page and cut and paste it into a file? Assume there isn't a unix (ideally) tool I could use to monitor this? Anything else?

thanks

Lee.

[Andy Burns]

grinch wrote:

> This is a simple  network traffic issue.

I'm not so sure, I think it's a problem on the copper loop to the home
causing bursts of packet errors.

[Graham J]

Lee Nowell wrote:
[snip]

>
> This thread seems to have split with a separate issue to my original one so hopefully I am replying to the correct bit of the thread. I spoke with EE earlier and they are sending out a Qube engineer on Thursday. Not sure what one of these is but are there any useful things I could monitor/ track between now and then which would help them troubleshoot the issue? Maybe how the results above change over time - assume I will need to do this manually on the router page and cut and paste it into a file? Assume there isn't a unix (ideally) tool I could use to monitor this? Anything else?

The Qube engineer may simply want to establish that there is nothing
wrong with your router and internal wiring, so he establishes that the
fault is real and can be escalated to Openreach. This is so EE can
avoid being charged an engineer callout fee by Openreach, which would
happen if Openreach determine that the fault is internal to your
property - the demarcation point being the master socket.

When I've seen this previously the engineer (sent by TalkTalk) was
Polish and spoke almost no English, so his visit was virtually
irellevant given that I had already established that the fault was not
internal to the property.

So what follows may be irrelevant. (Note that he includes she.)

1. The engineer will probably test from the test point within the master
socket, so you should already have tested from there so you know the
connection performance is not affected by wiring or phone equipment in
your house.

2. Be present all the time to see what he does.

3. Be helpful, volunteer tea/coffee/biscuits etc.

4. When he arrives, it would be par for the course to expect that he has
not been briefed as to the nature of the problem. Have some brief notes
typed up, including speed/noise figures and the times when you've noted
CRC and HEC error counts increasing.

5. Remember what the engineer does. Don't obviously make notes, but
show an interest without being critical or nosey. As soon as he leaves
write up what you remember.

6. It's likely he will disconnect your router and use his own for some
tests.

7. He may also have some line testing device (see if you can remember
the make - might be Hawk) which will look for reflections and other
transmission line parameters.


--
Graham J

[Lee Nowell]

Thanks both. Seems like this might not be as hopeful as I thought. We had an OpenReach engineer come a couple of weeks ago and sorted a constant disconnect/ reconnect issue we started having. He fixed it by redoing some of the joints between the house and the cabinet and then run a full test with the gadget he had and said all was well. If Qube are a precursor to another OpenReach engineer then doesn't sound like they are likely to find anything new. I was hoping they would be more of an end to end troubleshooter rather than the line itself.

As a side note, when EE run the automatic line tests earlier, they temporarily disconnected my broadband so I lost the historic DSL stats. Curiously it has now been up for 7 hours and has been pretty well behaved. Latest stats are.....

Status
Configured Current
Line Status -- UP

Link Type -- Fast Path

Operation Mode Automatic G.992.5 (ADSL2+)
Data Rate Information

Upstream 1247 (Kbps)
Downstream 14171 (Kbps)

Defect/Failure Indication
Operation Data Upstream Downstream

Noise Margin 6.5 (dB) 3.0 (dB)

Line Attenuation 15.0 (dB) 30.0 (dB)

Indicator Name Near End Indicator Far End Indicator

Output Power 12.4 (dBm) 0.0 (dBm)

Fast Path FEC Correction 0 0

Interleaved Path FEC Correction -- --

Fast Path CRC Error 682 1

Interleaved Path CRC Error -- --

Loss of Signal Defect 1 0
Fast Path HEC Error STR 1569 1

Interleaved Path HEC Error -- --

Error Seconds 8232 108

Statistics
Received Data 34359738 (Kbits)

Transmitted Data 22792562 (Kbits)

[Sn!pe]

grinch <gri...@somewhere.com> wrote:

[...]

> I don't have congestion issues with my Zen account but most of my home
> network is Cisco.I am on Zen FTTC with a working dual stack.
>

I too am on Zen FTTC. What does "working dual stack" mean please,
and what advantages does it confer?

--
^Ï^ <https://youtu.be/_kqytf31a8E>

My pet rock Gordon just is.

[Andy Burns]

Sn!pe wrote:

> What does "working dual stack" mean please,

IPv6 + IPv4

> and what advantages does it confer?

I think you can see an animated turtle somewhere :-)

[Sn!pe]

Ta! ≈:-)

[Graham J]

Lee Nowell wrote:

[snip]

I suspect EE also reset the DLM (Dynamic Line Management), hence the 3
dB SNR margin downstream. If there is any noise the DLM will adjust the
SNR margin upwards and reduce the speed, possibly only making these
changes around 2 am.

If your layout is correct most of the errors have occurred on the
upstream path. I suspect these are counted in the DSLAM and reported to
your router, so the counts may never be reset. I see a similar pattern
here and these counts are much the same across even a factory reset of
my router. So the question is whether these upstream counts increase
rapidly at the same times as you see the buffering.

You could ask the Qube engineer to explain why upstream errors are so
much higher than downstream.

--
Graham J

[grinch]

While that is a possibility it would cause more consistent traffic
issues. If you run winmtr you can get a rough idea as to where the
issues lie. You would need to have mtr running when you have issues

Dropped icmp packets usually mean a core router is busy as icmp is seen
as less important traffic by most core router makes.

[Lee Nowell]

I have been intermittently manually taking snapshots of the DSL stats over the past 2 days and put them in a spreadsheet and have uploaded it here

https://drive.google.com/file/d/1BGc3RNsxY6o-6TiqJ03_KqxNSbbXKMHh/view?usp=sharing

A few interesting points to note
1. The data received doesn't seem to change
2. The fast path errors seems to have steadily increased until the last one where it "shot up"
3. Error seconds seem to have started high but then gone up more slowly.

Does this tell us anything :) ?

Thanks again for your help

Lee.

[Graham J]

Lee Nowell wrote:

[snip]

> I have been intermittently manually taking snapshots of the DSL stats over the past 2 days and put them in a spreadsheet and have uploaded it here
>
> https://drive.google.com/file/d/1BGc3RNsxY6o-6TiqJ03_KqxNSbbXKMHh/view?usp=sharing
>
> A few interesting points to note
> 1. The data received doesn't seem to change
> 2. The fast path errors seems to have steadily increased until the last one where it "shot up"
> 3. Error seconds seem to have started high but then gone up more slowly.
>
> Does this tell us anything :) ?
>

1. Is probably a bug in the router's software.

2. Possibly a continuous background noise then some additional noise
which also increased the "Far End" counts.

3. Error Seconds was probably reset to zero at the last factory reset -
so potentially some time ago.

Did you have a thunderstorm or heavy rain during the time before the
last reading?

Worth showing this to your Qube engineer.

--
Graham J

[Andy Burns]

Lee Nowell wrote:

> https://drive.google.com/file/d/1BGc3RNsxY6o-6TiqJ03_KqxNSbbXKMHh/view?usp=sharing

>
> Does this tell us anything :) ?

Only that you've had about 10,000 errors (between the two levels of
correction) in just over a one hour window and they were bunched into
about 700 of the seconds within that hour.

Hasn't Mr Qube been yet?

[Lee Nowell]

Thanks both

@Andy - how did you deduce the numbers you mention? I should have mentioned that the first row gives the time I took the stats over the 2 days - should probably have given the date too to make it easier to see.

Qube due in the next few hours so hoping to get the stats/ questions all ready for them :)

[Martin Brown]

On 03/05/2021 21:53, Andy Burns wrote:
> Lee Nowell wrote:
>

>> Graham J wrote:
>>
>>> Can you get an upgrade to FTTC?
>>
>> I can although our cabinet is about 1/4 mile from the exchange
>
> With FTTC, what counts is distance from home to cabinet, exchange to
> cabinet is irrelevant, and the fibre probably goes to a bigger exxchange
> further away then the copper goes to anyway

And it is rather sensitive to distance. Much beyond 1200m from a cabinet
and you have to pray for a nice high quality newish wire (there are none
of those where I live). Our telecoms cabling is a selection of
ramshackle antique wiring nibbled by rodents and saturated with water.

>
> I'd say anyone on ADSL, who has the option of VDSL should jump at it,
> reliability and speed will increase, price difference is minimal.

Some in our village who can get VDSL have found it fast when it works
but much less tolerant of the decrepit local loop copper wiring in
flooded conduits with corroded connections. Water table is so high here
that they sometimes have the underground policeman's hat type "sealed"
junction units half full of water. It does nothing for the signal!

They have to let you switch back if VDSL doesn't work for you. I am a
case in point (apart from being on an EO line). If I went to VDSL it
would be 3.5km to the nearest FTTC box. It is 3km to the exchange.

They are allegedly bringing cable to my village (at least that is what
the guys digging up the street said). I am not holding my breath!

--
Regards,
Martin Brown

[Andy Burns]

Martin Brown wrote:

> Some in our village who can get VDSL have found it fast when it works
> but much less tolerant of the decrepit local loop copper wiring in
> flooded conduits with corroded connections.

I'm glad that I can /just/ see the FTTC cabinet from where I live, and
once I move I'll end-up even closer to a cabinet.

[Andy Burns]

Lee Nowell wrote:

> how did you deduce the numbers you mention?

yes, I just looked at first and last time columns, and assumed they were
on the same day.

[Lee Nowell]

Qube engineer has just left. He didn't seem too interested in my analysis or the figures. Only test device I saw was something that looked a bit like this but had a lot more buttons on it and don't recall it looking so much like a hand set

https://uk.rs-online.com/web/p/phone-line-testers/7659599/

He took a look at the master socket said it was a "star wiring" and an old fashioned "metal" one (not sure what was supposed to be metal given it is a normal master socket made of white plastic) and the issue is a build up of "magnetic" waves on the metal which makes it suddenly drop!! Anyway he has changed the master socket to a "modern" one and thinks that will solve the problem. Must say I am not holding my breath....

One thing he did do was move the master socket. The line comes into the house in the loft room where the master socket was located. From there I have some cat 6 wired in to the consumer side of the master socket which then takes the line to the other end of the house where the "comms room" is. This is then wired into a patch panel and the router connected to that. He crimped the incoming line to the cat 6 where the original master socket was and then fitted the new master socket in the "comms room" at the other end of the cat 6.

Feels a bit like a default answer of change the master socket to me.... Would appreciate your thoughts.

One thing he did say was that the incoming line comes from a telegraph pole and through a neighbour's tree to our house. He thought maybe the wind is blowing the tree and interfering with the incoming line.

[Andy Burns]

Lee Nowell wrote:
> On Thursday, 6 May 2021 at 10:49:03 UTC+1, Andy Burns wrote:
>> Lee Nowell wrote:
>>
>>> how did you deduce the numbers you mention?
>> yes, I just looked at first and last time columns, and assumed they were
>> on the same day.
>
> Qube engineer has just left. He didn't seem too interested in my analysis or the figures. Only test device I saw was something that looked a bit like this but had a lot more buttons on it and don't recall it looking so much like a hand set
>
> https://uk.rs-online.com/web/p/phone-line-testers/7659599/

so he only really looked at it as a "phone problem" not a "broadband
problem"

> He took a look at the master socket said it was a "star wiring" and
> an
> old fashioned "metal" one (not sure what was supposed to be metal given
> it is a normal master socket made of white plastic) and the issue is a
> build up of "magnetic" waves on the metal which makes it suddenly drop!!

sniff, sniff ...

> Anyway he has changed the master socket to a "modern" one and thinks
> that will solve the problem. Must say I am not holding my breath....>
>
> One thing he did do was move the master socket. The line comes into
> the house in the loft room where the master socket was located. From
> there I have some cat 6 wired in to the consumer side of the master
> socket which then takes the line to the other end of the house where the
> "comms room" is. This is then wired into a patch panel and the router
> connected to that. He crimped the incoming line to the cat 6 where the
> original master socket was and then fitted the new master socket in the
> "comms room" at the other end of the cat 6.

probably marginally better that way, but unlikely to fix a fault.

> Feels a bit like a default answer of change the master socket to me.... Would appreciate your thoughts.
>
> One thing he did say was that the incoming line comes from a
> telegraph
> pole and through a neighbour's tree to our house. He thought maybe the
> wind is blowing the tree and interfering with the incoming line.

Hopefully you've get to the point where they send a broadband engineer
next time ..

[Graham J]

Lee Nowell wrote:

>
> Qube engineer has just left. He didn't seem too interested in my analysis or the figures. Only test device I saw was something that looked a bit like this but had a lot more buttons on it and don't recall it looking so much like a hand set
>
> https://uk.rs-online.com/web/p/phone-line-testers/7659599/

That won't test anything to do with ADSL.

> He took a look at the master socket said it was a "star wiring"

That implies that there were several extensions each wired directly to
the master. Was that in fact true? Sounds like rubish from what
you've written later

> and an old fashioned "metal" one (not sure what was supposed to be metal given it is a normal master socket made of white plastic) and the issue is a build up of "magnetic" waves on the metal which makes it suddenly drop!! Anyway he has changed the master socket to a "modern" one and thinks that will solve the problem. Must say I am not holding my breath....

!!!!!!!!!!!!!!!!!!!!????????????????!!!!!!!!!!!!!!!!!!!

> One thing he did do was move the master socket. The line comes into the house in the loft room where the master socket was located. From there I have some cat 6 wired in to the consumer side of the master socket which then takes the line to the other end of the house where the "comms room" is. This is then wired into a patch panel and the router connected to that. He crimped the incoming line to the cat 6 where the original master socket was and then fitted the new master socket in the "comms room" at the other end of the cat 6.

That can't do any harm.

> One thing he did say was that the incoming line comes from a telegraph pole and through a neighbour's tree to our house. He thought maybe the wind is blowing the tree and interfering with the incoming line.

If it's actually chafing the line that could cause your problems. Can
you actually see this for yourself? Does the poor ping symptom occur at
the same time as wind is moving the tree against the line?

If you could surreptitiously break the line by poking a long stick at it
to make it appear that it was broken by the tree, then your voice
service provider (which might not be EE, so do tell us) will have to
send Openreach to fix it. Openreach may be prepared to trim the tree to
prevent further damage. Or they might run your line from a different
pole ...


--
Graham J

[Lee Nowell]

On Thursday, 6 May 2021 at 15:09:49 UTC+1, Graham J wrote:
> Lee Nowell wrote:
>
> >
> > Qube engineer has just left. He didn't seem too interested in my analysis or the figures. Only test device I saw was something that looked a bit like this but had a lot more buttons on it and don't recall it looking so much like a hand set
> >
> > https://uk.rs-online.com/web/p/phone-line-testers/7659599/
> That won't test anything to do with ADSL.
> > He took a look at the master socket said it was a "star wiring"
> That implies that there were several extensions each wired directly to
> the master. Was that in fact true? Sounds like rubish from what
> you've written later

No was not true. Essentially Master socket -> patch panel -> ADSL filter which then splits to Router and 1 phone line

> > and an old fashioned "metal" one (not sure what was supposed to be metal given it is a normal master socket made of white plastic) and the issue is a build up of "magnetic" waves on the metal which makes it suddenly drop!! Anyway he has changed the master socket to a "modern" one and thinks that will solve the problem. Must say I am not holding my breath....
> !!!!!!!!!!!!!!!!!!!!????????????????!!!!!!!!!!!!!!!!!!!
> > One thing he did do was move the master socket. The line comes into the house in the loft room where the master socket was located. From there I have some cat 6 wired in to the consumer side of the master socket which then takes the line to the other end of the house where the "comms room" is. This is then wired into a patch panel and the router connected to that. He crimped the incoming line to the cat 6 where the original master socket was and then fitted the new master socket in the "comms room" at the other end of the cat 6.
> That can't do any harm.
> > One thing he did say was that the incoming line comes from a telegraph pole and through a neighbour's tree to our house. He thought maybe the wind is blowing the tree and interfering with the incoming line.
> If it's actually chafing the line that could cause your problems. Can
> you actually see this for yourself? Does the poor ping symptom occur at
> the same time as wind is moving the tree against the line?
>
> If you could surreptitiously break the line by poking a long stick at it
> to make it appear that it was broken by the tree, then your voice
> service provider (which might not be EE, so do tell us) will have to
> send Openreach to fix it. Openreach may be prepared to trim the tree to
> prevent further damage. Or they might run your line from a different
> pole ...

Yes voice is also with EE. Good idea - the tree is my neighbours but I might be able to get a ladder up to take a look and see if there are any signs of issue. Just finished work so setting up my tests again. Today there is no wind at all so will be an interesting test from this perspective

>
>
> --
> Graham J

[NY]

"Lee Nowell" <leen...@yahoo.co.uk> wrote in message
news:5b76a17f-22cf-4c29...@googlegroups.com...

>> > One thing he did say was that the incoming line comes from a telegraph
>> > pole and through a neighbour's tree to our house. He thought maybe the
>> > wind is blowing the tree and interfering with the incoming line.
>> If it's actually chafing the line that could cause your problems. Can
>> you actually see this for yourself? Does the poor ping symptom occur at
>> the same time as wind is moving the tree against the line?
>>
>> If you could surreptitiously break the line by poking a long stick at it
>> to make it appear that it was broken by the tree, then your voice
>> service provider (which might not be EE, so do tell us) will have to
>> send Openreach to fix it. Openreach may be prepared to trim the tree to
>> prevent further damage. Or they might run your line from a different
>> pole ...
>
> Yes voice is also with EE. Good idea - the tree is my neighbours but I
> might be able to get a ladder up to take a look and see if there are any
> signs of issue. Just finished work so setting up my tests again. Today
> there is no wind at all so will be an interesting test from this
> perspective

I remember back in the 70s we had tremendous problems with crackling on
phone calls. We had numerous GPO engineers to investigate. They changed the
carbon mike in the telephone and then the cable from the pole to the house
(which didn't go via any trees!). They changed the house wiring. I forget
what finally fixed it, but what I remember is that the engineers gave me the
old parts (which would only be thrown away) so I had a carbon mike and about
100 feet of very stiff copper two-core wire. I was hoping that he'd decide
that the whole phone needed to be changed, and I might acquire the old one,
but that didn't happen! I remember using the mike (in conjunction with a
parabolic metal photography light) to make a very sensitive directional mike
that would pick up conversations at the bottom of the garden; and I had a
long cable from my bedroom to the garden shed so I could play music from my
record player on a speaker (found in a skip at the local tip*) in the shed.


(*) Back in the days when the council just left a couple of skips in a layby
and came back to collect them the following week. Everyone took their stuff
there, and there were always people rooting in the skips to see what they
could find, with no council staff there to prohibit this.

[Graham J]

Lee Nowell wrote:

[snip]

>
> Yes voice is also with EE. Good idea - the tree is my neighbours but I might be able to get a ladder up to take a look and see if there are any signs of issue. Just finished work so setting up my tests again. Today there is no wind at all so will be an interesting test from this perspective

Does the cable actually pass over your neighbour's land? If so maybe
Openreach has to pay your naighbour for a wayleave?

Or does his tree extend over your land or the public highway? If so
Openreach can probably trim it with impunity.


--
Graham J

[Martin Brown]

I don't know why it is but on a marginal unreliable line I also see
upstream hard error seconds accumulating at a 10:1 ratio to downstream
as well. >100:1 on correctable errors. May be a feature of ADSL 2+.

> I have been intermittently manually taking snapshots of the DSL stats over the past 2 days and put them in a spreadsheet and have uploaded it here
>
> https://drive.google.com/file/d/1BGc3RNsxY6o-6TiqJ03_KqxNSbbXKMHh/view?usp=sharing
>
> A few interesting points to note
> 1. The data received doesn't seem to change

It has hit some random constant value. See my other post about it.

Mine is stuck at the same value. I took a very close lightning strike
this afternoon - one where I saw a fat spark jump from the lead to the
modem! I was amazed when it rebooted OK after mains power was restored
to the village.

This enormous spike of current/voltage seems to have removed some oxide
since I now the best download sync speed I have ever seen at 8.071Mbps.

> 2. The fast path errors seems to have steadily increased until the last one where it "shot up"

Ask them to put you one Interleaved unless you are a gamer.

> 3. Error seconds seem to have started high but then gone up more slowly.

Probably that their attempts to use a TDR pulse measurement to test for
a dry joint have partially healed the fault for now by breaking down the
metal oxide coat.

> Does this tell us anything :) ?

Forcing a power reset on your modem will restart the statistics from the
baseline (just don't do it too often). Right now you are blind to your
error rate since the download counter has maxxed out at its random magic
number. FWIW Mine also maxes out at the same magic number and I have a
stable enough connect that it spends a fair amount of time pegged out.

--
Regards,
Martin Brown

[Roderick Stewart]

On Thu, 6 May 2021 20:26:50 +0100, "NY" <m...@privacy.invalid> wrote:

>(*) Back in the days when the council just left a couple of skips in a layby
>and came back to collect them the following week. Everyone took their stuff
>there, and there were always people rooting in the skips to see what they
>could find, with no council staff there to prohibit this.

I remember trudging over a local landfill site and collecting all
manner of useful stuff, including several complete radios, sometimes
in working order or with not much wrong with them. Sadly, today's
youngsters don't seem to have anything like the same learning
opportunities that I had.

Rod.

[Martin Brown]

On 06/05/2021 13:42, Lee Nowell wrote:
> On Thursday, 6 May 2021 at 10:49:03 UTC+1, Andy Burns wrote:
>> Lee Nowell wrote:
>>
>>> how did you deduce the numbers you mention?
>> yes, I just looked at first and last time columns, and assumed they
>> were on the same day.
>
> Qube engineer has just left. He didn't seem too interested in my
> analysis or the figures. Only test device I saw was something that
> looked a bit like this but had a lot more buttons on it and don't
> recall it looking so much like a hand set
>
> https://uk.rs-online.com/web/p/phone-line-testers/7659599/
>
> He took a look at the master socket said it was a "star wiring" and
> an old fashioned "metal" one (not sure what was supposed to be metal
> given it is a normal master socket made of white plastic) and the
> issue is a build up of "magnetic" waves on the metal which makes it
> suddenly drop!! Anyway he has changed the master socket to a
> "modern" one and thinks that will solve the problem. Must say I am
> not holding my breath....

Assuming that you mean one with the bell wire hack implemented then it
might be good for a +10% speed improvement. I got a bit more than that.

> One thing he did do was move the master socket. The line comes into
> the house in the loft room where the master socket was located. From
> there I have some cat 6 wired in to the consumer side of the master
> socket which then takes the line to the other end of the house where
> the "comms room" is. This is then wired into a patch panel and the
> router connected to that. He crimped the incoming line to the cat 6
> where the original master socket was and then fitted the new master
> socket in the "comms room" at the other end of the cat 6.

I have tried with and without 30m of additional cabling in my circuit
and could not tell the difference. Such lengths are negligible compared
to the distance to the exchange provided that you don't run it parallel
to a noisy mains cable or something daft like that or have bad joints.

> Feels a bit like a default answer of change the master socket to
> me.... Would appreciate your thoughts.

It is a reasonable thing to try. It should also mean you have a clean
test socket somewhere that eliminates all house wiring.

> One thing he did say was that the incoming line comes from a
> telegraph pole and through a neighbour's tree to our house. He
> thought maybe the wind is blowing the tree and interfering with the
> incoming line.

That is bad news. You will get microphonics from the tree branches
rubbing the cable and eventually loud clicks and bangs are the
insulation on the drop wire begins to fail. It is quite common here. We
have a lot of trees. Electricity company cut them back every couple of
years - I have *never* seen BT do it though. Cables run along roughly
the same poles so most times it works but not always.

BTW one trick you can do to fine tune your sync speed is deliberately
force a resync when the SNR is at its worst which is around midnight (at
least 3 hours after sunset). Or if you want to force the fastest sync an
hour or two after midday. The solar wind affects the ionosphere and
propagation of MW radio from the continent which can in interfere with
ADSL/VDSL signal to noise. Certainly when doing comparative tests you
should try to resync at the same time of day to eliminate this natural
diurnal variation from the equation.

--
Regards,
Martin Brown