Print

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

[burakkucat]

A little earlier this evening I called up MDWS to check the status of Kitz' circuit . . . and it was not a happy sight. 

The oscillations in the SNRM are back with a vengeance and the circuit had collected 11107 error seconds (DS) in 23 hours.

[Chrysalis]

big ouch, I am guessing something is interfering on the line, probably electrical.

That is one hell of an annoying problem to have, you basically have a line that most days can hold fast path.
You get an occasional maybe once every 2-4 weeks glitch.
DLM then acts delayed hours after the event and then you stuck on a gimped profile for 1-2 weeks as a result, all for a one glitch.

An engineer turning up when the glitch isnt happening will have a hard time finding fault also.

So was about 4 hours duration.

[burakkucat]

big ouch, I am guessing something is interfering on the line, probably electrical.

Yes, I would agree. The problem is trying to identify exactly what . . . 

[Chrysalis]

When I first got FTTC and still on BT retail, in the first year of service a problem developed where randomly during the night I would get a load of packet loss that lasted for a short while.  It would disappear as fast as it appeared.

I cannot remember all the specific's but what I do remember is this.

It was service affecting hence the term "packet loss", and if I remember right CRC errors did not accumulate on the line, so it was like some kind of invisible problem on the line not registering on the stats, snrm didnt fluctuate etc.

An engineer came, said he did some checks and found nothing wrong but then the problem mysteriously never came back after the engineer visit.

[burakkucat]

I wonder what (visible) effect SRA would have on an SNRM plot? 

[Chrysalis]

it would make the SNRM flat, and the sync speed itself would vary instead.

Here is a sync speed and snrm graph from my ukonline days.

You can see around 7am the sync speed suddenly dips, without SRA that would make my line lose sync or at the very least cause massive packet loss due to excessive error rates, but with SRA the sync speed temporarily dropped to maintain a healthy snrm and increased again when the issue ended.

My adsl line used to get frequent issues during weekday daylight hours starting from around 7am.

You can also see the sync speed adjustment between day and night on there as well.

[burakkucat]

Thank you for sharing those historic plots and the accompanying explanation.

All is now clear.

[Chrysalis]

no problem

[Chrysalis]

back on fast path but that SNRM looks a complete mess. Was steady for a bit on the 4th before going crazy again at midnight.  That tells me its not the line but some kind of interference.

[burakkucat]

Yes, I would tend to agree. 

[PhilipD]

Hi

Looks a lot better this morning, something happened at around 5:40am this morning, now looks good.

Regards

Phil

[burakkucat]

Yes, things are now looking more normal.

Whatever was causing the problem, it abruptly cleared without any need for a CPE - DSLAM resynchronisation event. I will continue to monitor the circuit . . .

[PhilipD]

Hi

Looks like the noise problem is back from around 00:30 on the 27 Feb.  I wonder if someone was away for a weeks holiday and arrived back early hours on the 27 Feb and turned back on whatever it is causing the noise?

Regards

Phil

[kitz]

Yes neighbour was away for half term.   However the dates & times dont exactly fit - almost but not quite.

Line became ok at about 05:30 on Tue 21st Feb
Problem started again at about 00:35 on Mon 27th Feb

She was away from Sat 18th of Feb until Sun 26th Feb.

[PhilipD]

Hi

Would have been convenient if it was them to sort it out.  Could be anyone else of course, and some faulty power supply to a router could be injecting noise back down the line to be picked up by lines close by.  It does suspiciously look like a person/property is responsible and they went away turning off the equipment causing the problem.

Regards

Phil