Print

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

[ColinS]

There is a thread elsewhere here on Kitz that discusses, or at least speculates upon, some aspects of the DLM algorithm.  If DLM has seen resyncs to be the issue (rather than error rates), then it will view the problem as a primarily one of stability.  In those circumstances, 'how long it takes' depends upon whether, having 'temporarily' capped your sync rates at a lower profile, it encounters any further 'instability'.  There is some evidence to suggest that 14 days is a basic unit of the time which the algorithm uses.  Unfortunately it also uses a 'doubler' effect each time further instability is encountered while already in a capped profile.
So, for example only, a one off lowering of the profile might be lifted in 14 days, but (as happened in my case), if this happened 3 times in a row (as it did), it took closer to 2^(3-1)*14=~56 days to return to 'normal', albeit that it did progressively step up the profiles during that time.
I only mention this to give you some idea of the timescales that could be involved - they can test the EU's patience considerably. HTH

[DeadMan]

As suspected not DLM reset. I bet he did not even try. On the phone right now. Apparently I am on/was a 35 meg banded profile (bit of confusion about that actually as he then said later on it is 60). They are waiting for the engineer notes to come through but he is going to try to get me off the banded profile if he can and look further into what can be done. I explained that the engineer was a bit clueless (probably fresh out of training he seemed young) and would not entertain any of my suggestions like a pair swap etc (I think OR are coming down tough on engineers to do as little as possible. Either that or it was because he was clueless or worse, lazy) The device they use will only give a snapshot of line conditions for those few minutes it is on so I don't really think they are worth a carrot for time based issues.

According to the support guy they can take you off banding profile and put you on fastpath permanently? I thought only the DLM had control of that or does it not? I asked to be set on an unmanaged  profile if possible.

Anyhow I now have to wait for the engineers notes to come through to see what the next step is.

[DeadMan]

Still waiting for something to be done. They said they changed my banding which increased speed tests but not sync rate. My interleave is now 600 on upload. I foolishly did a resync earlier today and since then my speed tests are bouncing up and down from 44mbps-57mbps and my attainable rate is fluctuating very slightly where as before it was always mostly a smooth line.

Plusnet asked to do another BTW speed test but it does not tell the entire story. I do not want them coming back to me again with the 'within estimates' excuse. If they say they cannot fix the line then so be it. But at least get someone to reset it all to defaults from my first install so I can start from a fresh baseline to measure from.

[DeadMan]

One thing I forgot to mention is that the master socket is NOT the NTE5 socket. The master socket is a very old smaller faceplate/box that runs a worn looking cable going out a hole drilled into the window pane. The NTE5 box is directly above it so I presume the cabling inside the master socket is then connected via another bit of wiring to the NTE5 faceplate. I know this because in the past I have had the master socket off and know the cable coming into the house is too short to extend to the NTE5 faceplate without an extension.

[mikehiow]

There is a thread elsewhere here on Kitz that discusses, or at least speculates upon, some aspects of the DLM algorithm.  If DLM has seen resyncs to be the issue (rather than error rates), then it will view the problem as a primarily one of stability.  In those circumstances, 'how long it takes' depends upon whether, having 'temporarily' capped your sync rates at a lower profile, it encounters any further 'instability'.  There is some evidence to suggest that 14 days is a basic unit of the time which the algorithm uses.  Unfortunately it also uses a 'doubler' effect each time further instability is encountered while already in a capped profile.
So, for example only, a one off lowering of the profile might be lifted in 14 days, but (as happened in my case), if this happened 3 times in a row (as it did), it took closer to 2^(3-1)*14=~56 days to return to 'normal', albeit that it did progressively step up the profiles during that time.
I only mention this to give you some idea of the timescales that could be involved - they can test the EU's patience considerably. HTH

It's worth mentioning that having been banded from 80/20 to 60/20, since resolving the issue that caused me to be banded in the first place, I was back on 73/20 within 2 days and back on 80/20 within 4.

[ColinS]

It's worth mentioning that having been banded from 80/20 to 60/20, since resolving the issue that caused me to be banded in the first place, I was back on 73/20 within 2 days and back on 80/20 within 4.

Sure.  That can (and obviously does) happen as long as there are no further instances of whatever (instability) caused it to be capped in the first place.  Otherwise e.g. if further resyncs are caused while you are already on a capped profile, then the delay doubling mechanism may kick in.  In my case my own line stepped back up two profiles in exactly the same time frame as happened to you, but then took a further 2^(3-1)*14days to return to the full 80/20.
So it all depends upon each individual set of circumstances, including: the nature of the fault, its duration, whether it recurs on subseqent days, the DSLAM type you're on, how many profile bands you have been stepped down, etc.
But certainly, with a fair wind, DLM can also recover quite quickly.  The problems really occur when it doesn't.

[burakkucat]

One thing I forgot to mention is that the master socket is NOT the NTE5 socket. The master socket is a very old smaller faceplate/box that runs a worn looking cable going out a hole drilled into the window pane. The NTE5 box is directly above it so I presume the cabling inside the master socket is then connected via another bit of wiring to the NTE5 faceplate. I know this because in the past I have had the master socket off and know the cable coming into the house is too short to extend to the NTE5 faceplate without an extension.

   Hmm . . . I find that a little difficult to follow. Could you possibly take a series of photographs and make them available for viewing, please?

[DeadMan]

Todays stats. Something happened at 8am after a resync (I was in bed so not me!)

# xdslcmd info --stats
xdslcmd: ADSL driver and PHY status
Status: Showtime
Retrain Reason: 2
Max: Upstream rate = 18534 Kbps, Downstream rate = 75604 Kbps
Path: 0, Upstream rate = 20000 Kbps, Downstream rate = 65416 Kbps

Link Power State: L0
Mode: VDSL2 Annex B
VDSL2 Profile: Profile 17a
TPS-TC: PTM Mode
Trellis: U:ON /D:ON
Line Status: No Defect
Training Status: Showtime
Down Up
SNR (dB): 6.1 4.9
Attn(dB): 0.0 0.0
Pwr(dBm): 13.3 6.8
VDSL2 framing
Path 0
B: 51 237
M: 1 1
T: 64 45
R: 12 16
S: 0.0253 0.3782
L: 20243 5373
D: 1281 1
I: 64 127
N: 64 254
Counters
Path 0
OHF: 15507772 1980373
OHFErr: 143 101
RS: 3969833459 243514
RSCorr: 407619 557
RSUnCorr: 9285 0

Path 0
HEC: 2070 0
OCD: 58 0
LCD: 58 0
Total Cells: 3169782522 0
Data Cells: 8575543 0
Drop Cells: 0
Bit Errors: 0 0

ES: 115 79
SES: 11 0
UAS: 77 77
AS: 25202

Path 0
INP: 3.00 0.00
PER: 1.61 4.25
delay: 8.00 0.00
OR: 118.61 60.17

Bitswap: 3584 618

Total time = 1 days 12 hours 25 min 37 sec
FEC: 1902224 169402
CRC: 6712 106
ES: 115 79
SES: 11 0
UAS: 77 77
LOS: 5 0
LOF: 10 0
Latest 15 minutes time = 10 min 37 sec
FEC: 5267 12
CRC: 0 2
ES: 0 2
SES: 0 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
Previous 15 minutes time = 15 min 0 sec
FEC: 23075 9
CRC: 5 2
ES: 1 2
SES: 0 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
Latest 1 day time = 12 hours 25 min 37 sec
FEC: 658936 26884
CRC: 3185 101
ES: 47 75
SES: 5 0
UAS: 20 20
LOS: 5 0
LOF: 5 0
Previous 1 day time = 24 hours 0 sec
FEC: 1243288 142518
CRC: 3527 5
ES: 68 4
SES: 6 0
UAS: 57 57
LOS: 0 0
LOF: 5 0
Since Link time = 7 hours 1 sec
FEC: 407619 557
CRC: 143 101
ES: 26 75
SES: 0 0
UAS: 0 0
LOS: 0 0
LOF: 0 0
#

# xdslcmd info --pbParams
xdslcmd: ADSL driver and PHY status
Status: Showtime
Retrain Reason: 2
Max: Upstream rate = 18485 Kbps, Downstream rate = 75604 Kbps
Path: 0, Upstream rate = 20000 Kbps, Downstream rate = 65416 Kbps

Discovery Phase (Initial) Band Plan
US: (0,95) (868,1207) (1972,2783)
DS: (32,859) (1216,1963) (2792,3959)
Medley Phase (Final) Band Plan
US: (0,95) (868,1207) (1972,2783)
DS: (32,859) (1216,1963) (2792,3959)
VDSL Port Details Upstream Downstream
Attainable Net Data Rate: 18485 kbps 75604 kbps
Actual Aggregate Tx Power: 6.8 dBm 13.3 dBm
============================================================================
VDSL Band Status U0 U1 U2 U3 D1 D2 D3
Line Attenuation(dB): 4.3 23.6 36.0 N/A 11.8 29.1 45.0
Signal Attenuation(dB): 4.3 23.2 34.9 N/A 11.8 29.1 45.0
SNR Margin(dB): 5.0 5.0 4.9 N/A 6.1 6.1 6.1
TX Power(dBm): -4.1 -19.5 6.4 N/A 10.1 7.9 7.0
#

[burakkucat]

Thank you for those pictures. I now understand.

Hopefully the original LJU master socket is not actually connected into the circuit and is just being used as means to hold gel-crimps which extend the service cable into the NTE5/A + SSFP assembly. 

[DeadMan]

Thank you for those pictures. I now understand.

Hopefully the original LJU master socket is not actually connected into the circuit and is just being used as means to hold gel-crimps which extend the service cable into the NTE5/A + SSFP assembly. 

I think so. I will take the master plate off to verify but pretty sure it is.

[DeadMan]

Had another BT SFI. Same old same old. BTO must really be drumming it into BT engineers heads to do as little as possible and NOT to touch the DLM unless absolutely necessary. Blood from a stone comes to mind.

I asked him if it was possible to removed the NTE5 box from the window frame and make it the master socket instead of the little socket underneath. I stressed to him that I did not wish to be charged otherwise leave it where it is and I will rewire from where it meets the back plate of the master socket to the NTE5 box myself. After asking him several times that I would not be charged for putting the NTE5 box under the internal window ledge with a small extension he put a small junction box with the master socket was. Removed the NTE5 box and put it under the ledge stuck to the wall with some double sided (it's loose and not screwed to the wall but I am not bothered too much as it does not have anywhere to fall other than dangle off the bit off cable. I will get some of that 'no more nails' stuff if it does.

Re-synced at lower speeds because it's the period of the day when SNRM is 1.5dB lower. Posted this info along with a little bit of a moan to Plusnet support and a BTW speedtest. If BT try to charge me they can jump off a cliff! It's something I could have done myself for free otherwise and I told the engineer this is what I had intended to do if he could not do it for free (you can see I am a bit worried about getting charged for this heh).

Anyhow I am back to square one and it looks like nobody will do anything as I am above speed estimates.

[Black Sheep]

"Had another BT SFI. Same old same old. BTO must really be drumming it into BT engineers heads to do as little as possible and NOT to touch the DLM unless absolutely necessary. Blood from a stone comes to mind".

For info -- Performing a DLM 'Re-calc' for an engineer is quite easy. We have 3 ways in which we can achieve this ..... 1) Via the WHOOSH GEA platform, accessed personally from the engineers laptop. 2) Via their mobile phone using our 'Call platform services'. 3) Via a remote operator, who has to ask the question, 'was any fault was found' ?.

We are not monitored on performing DLM 're-calcs', nor does it form any part of the 'Performance Management' statistics, or subsequent disciplinary procedures. So, in a nutshell, carrying out the deed is relatively simple.

However, you have hit the nail on the head in terms of not doing a 're-calc', unless necessary. By that, I would expect the engineer to have either located and fixed a fault, or, can prove from historic data that DLM has been overly-aggressive for whatever reason. Continued and constant 're-calc' of a circuit does not achieve anything in the long term.

[DeadMan]

OK it fell off so I put no more nails stickies on the back plate. Thanks for that titbit of info. Seems like he was just too lazy himself to do it. It would not have taken much it would seem for him to do it. Granted it would probably over time just go back to similar speeds. I was more curious as to what the initial sync speeds were when it was first installed without DLM doing anything to be honest (I do not know what it was with the ECI which was originally provided and can only guess at it).

I still say DLM sucks. But I am greedy lol.