Router, Line or the will of BT?

[#1]

Two/three thoughts. Have you got any other phone sockets at all apart from this bog standard BT Faceplate? I assume it looks like the middle one at http://www.plus.net/support/broadband/master-socket-guide.shtml
Are you able to trace the cable from that socket to the underground connection or telegraph pole?
Have you checked you phone bill lately to make sure there is no unusual phone call activity?

[#2]

From SamKnows it states
You are approximately 461 metres from the exchange. Note that this is the straight line distance - the actual cable length will be longer!
The 1 km is my estimate on the line length, with local geographical knowledge.
Theres only a standard corded phone in the same socket, closest match is the middle picture, the LJU. It and the router are filtered (brand new filter) and previous tests eliminated the phone as a source. As far as I know the DSLAM I'm connected to is manufactured by TI, I don't know if this has any bearing on the issue though.

[#3]

I'd be following that cabling to make sure it doesn't go anywhere it shouldn't or that there's no damage to it anywhere, and look for any joint boxes, if outside, are they keeping the rain out?

[#4]

Just done another BT Speedtest and the profile is now at 7.15  
I'd appreciate a PN Profile bump
cheers
Graham

[#5]

The plot thickens...
I left the router up for around an hour after the SNRM reset, but still had wildly fluctuating SNRM stats although the router was reporting zero errors and the connection remained up and stable. After the hour I switched the router off and left it off for around 90 minutes, then switched it back on again. See attached graph from routerstats.
Sync is 8128 Kbps, SNRM is 10.8 db and constant.
My theory is an issue between the Broadcom chipset in the router and the Texas Instruments DSLAM/Line Card at the exchange.
cheers
Graham

[#6]

Ah, arrh, not sure about the theory, but have seen and experienced this before but not as wild a variation, 1st graph noise variation present, 2nd it's not - a more typical day, but not the quietest.
I know they are different days & sync speeds but I'd have to go to another machine to find 2 consecutive examples on the same day, they are just to illustrate the point. Sometimes if I have that noise a resync gets rid of it, other times not and I have to wait several days. I've always put the noise down to some neighbours equipment, possibly next doors sky box or a nearby modem/router kicking out some cr*p and sometimes at the point of sync the DLM can't quite get around the correct bits/tone allocations if it's a bit severe, but your theory may be equally valid, but it's undoubtedly caused by whatever background noise is there at the time.
By the way what sample rate are you using, mine is set to 8 seconds.
I don't suppose you got the bits/tone plots when the noise variation was there and just after resync when it wasn't?
Edit: PS I think you should edit the title and make the last bit "will of DLM" rather than BT 

[#7]

Sample rate is set at 10 seconds, since the restart the noise line is flat at 10.8 db.
I've never seen the earlier behaviour with a resync before, as I've always had a completely stable line. The fly in the ointment this time was a router that had given up the struggleand took early retirement.
I shall keep the monitoring going overnight as I'd like to see the results with a lower SNRM during the period I've had most trouble with, around 3-4am.
cheers
Graham

[#8]

Hi Graham,
I've set your profile to 7150 so you should see the difference shortly.
Jojo

[#9]

Thanks Jojo
all looking good so far 
cheers
Graham

[#10]

A flat line 10.8 looks very dubious to me!
A 0.5db variation is totally normal and yours doesn't look a lot more than that. Only if the variation was associated with errors would I worry.

[#11]

He hasn't got 0.5 db that's several dB's.  0.5dB    no way    0.1dB maybe, but why not a flat line - he lives in the exchange (well 1km away, underground feed as well probably)

[#12]

After the SNRM reset, the line was showing stats varying between 7db and 10db, so overall a doubling (or halving depending on your point of view) of noise. This lasted for well over an hour, see graph in post #34  
Only after powering off the router for 90 minutes did the SNRM stabalise.
I still believe this is some issue between the modem/router and the DSLAM more of which I'll post in a moment.
Edit: Yep anotherone, the feed is all underground from the exchange to the cabinet and cabinet to the house. As the cable comes to the house it is in a fixed conduit up the wall for about 18 inches before entering.
cheers
Graham

[#13]

OK, here's some further details, I've been trying to gather enough tech knowledge so it appears I know what I'm talking about but I'm stumped with a few things at the moment. Any help understanding this information will be greatly appreciated.
First off, since Jojo reset the Target SNRM yesterday and letting the line settle, I've had a very stable connection overnight, see first graph. This covers the time period from 0130 to 0430 this am and is the time period where I had seen significant problems earlier.
This continued until 0911 this morning when I noticed the router was resynching and at first I could see no reason for this. Investigating further with the aid of routerstats and the syslog which were both running since yesterday, there were few indications as to why the router resynched.
There was a slight change in Rx Noise margin, was 10.8 - 10.9 db, now 11.0 to 11.1 db
There was a change to the TX Noise margin from 21db to 22 db
No change in Rx/Tx sync rates 8128/448
A number of errors were generated during the resync but I think I can safely ignore those.
Other than this, the only other spike was a burst of Bitswapping activity which occured at the same time as the resync, see second graph
Just to be sure, I'd also been logging the Bits & Attenuation graph in routerstats. Attached is an animated GIF which shows the difference between the Tones before and after the reync.

I can't see if this was the cause or the effect, but it seems like the bits are being migrated further up the spectrum, both Tx and Rx.
For completeness, heres the latest stats:

adsl: ADSL driver and PHY status
Status: Showtime
Retrain Reason: 1
Last initialization procedure status:   0
Max:     Upstream rate = 1012 Kbps, Downstream rate = 8768 Kbps
Channel: FAST, Upstream rate = 448 Kbps, Downstream rate = 8128 Kbps
Link Power State:       L0
Mode:                   G.DMT 
TPS-TC:                 ATM Mode
Trellis:                ON
Line Status:            No Defect
Training Status:        Showtime
                Down            Up
SNR (dB):       11.0            22.0
Attn(dB):       12.5            10.5
Pwr(dBm):       19.8            11.9
K:              255(0)          15
R:              0               0
S:              1.0000          1.0000
D:              1               1
SF:             737145          737121
SFErr:          0               0
RS:             0               0
RSCorr:         0               0
RSUnCorr:       0               0
HEC:            0               0
OCD:            0               0
LCD:            0               0
Total Cells:    240226118       0
Data Cells:     1696842         0
Drop Cells:     0
Bit Errors:     0               0
ES:             10              0
SES:            10              0
UAS:            25              25
AS:             12532
INP:            0.00            0.00
INPRein:        0               0
delay:          0               0
PER:            0.00            0.00
OR:             32.00           32.00
Bitswap:        0/0             0/0
Total time = 16 hours 37 min 10 sec
FEC:            0               0
CRC:            428             0
ES:             10              0
SES:            10              0
UAS:            25              25
LOS:            1               0
LOF:            9               0
LOM:            0               0
Latest 15 minutes time = 7 min 10 sec
FEC:            0               0
CRC:            0               0
ES:             0               0
SES:            0               0
UAS:            0               0
LOS:            0               0
LOF:            0               0
LOM:            0               0
Previous 15 minutes time = 15 min 0 sec
FEC:            0               0
CRC:            0               0
ES:             0               0
SES:            0               0
UAS:            0               0
LOS:            0               0
LOF:            0               0
LOM:            0               0
Latest 1 day time = 16 hours 37 min 10 sec
FEC:            0               0
CRC:            428             0
ES:             10              0
SES:            10              0
UAS:            25              25
LOS:            1               0
LOF:            9               0
LOM:            0               0
Previous 1 day time = 0 sec
FEC:            0               0
CRC:            0               0
ES:             0               0
SES:            0               0
UAS:            0               0
LOS:            0               0
LOF:            0               0
LOM:            0               0
Since Link time = 3 hours 28 min 51 sec
FEC:            0               0
CRC:            0               0
ES:             0               0
SES:            0               0
UAS:            0               0
LOS:            0               0
LOF:            0               0
LOM:            0               0

16 hours 37 min 10 sec equtes to the time I switched the router back on last night..
3 hours 28 min 51 sec equtes to the unsolicited resync at 0911 this morning.
The errors shown all occured within the 10 seconds it took to resync, so I think this is normal. However the one item in the log which may shed some light is Retrain Reason: 1 however I cannot find any information about this snippet from the log.
cheers
Graham

[#14]

Just to add to the thread that I've seen the same variation as poster "Anotherone"
Indeed my graphs were very similar.
I changed routers/ changed chipsets, checked on the master test socket etc - it still did it.
As a result of totally unconnected faults and cable thefts much of my line has over the last X years been changed.
Still this variation.
It was almost as if I got either a "good line" or a "bad line" at router switch on - silly of course as the physical line is exactly the same.
I came to the conclusion that it was the DSALM at the exchange that was the problem.
If I left the router on 24/7 when it was "good" then at some time (days/weeks)  would suddenly swap to 'bad'.
The solution was to switch off each night until upon switch on in the morning it reverted to "good".
To be honest I forgot about it and accepted that that was how things were.
Significantly I was recently swapped to 21CN/ADSL2+
and interestingly so far the daytime trace everyday has been totally flat +/- 0.1 dB on the SNR margin -which is on a 3dB target
I await so see what happens over the next few months but I'm tending to the view that it was the 20CN DSLAM all along.

[#15]

"Internet is working fine today" 


Apart from the one resync yesterday (I presume as part of the line training) all has been fine, speeds are up and ping is less than 50% of it's previous values.