drop outs in the small hours are likely DLM doing its thing. speeding you up a tiny bit if the line is performing well and can go faster without drops. slowing you down a tiny bit if its not and dropping out. DLM changing the speed causes drop and resync. its part of how your line is managed.

the electro magnetic interference on your line is not fixed. it changes pretty much constantly. as the cable goes off down the road to the cab and dslam it'll pick up all sorts of stuff from outside sources and even other lines alongside it. so a house several doors up the road unwittingly injecting a load of em noise onto their cable and then everyone else's in the bundle for example. winter wet/damp and wind is good for impacting all the many joints and overhead bits of cable. dsl has to 'deal' with all this rubbish as best it can while keeping you connected. that copper is a great antenna for stuff you don't want.

short bursts of noise and dsl will count up some errored seconds. too many and dsl will drop and you'll get a resync. DLM is watching those stats and adjusting. so as you said, lots of em from seasonal xmas lights etc, it probably slowed your sync rate a bit. now they're starting to get packed away, it'll speed you up again hopefully. your router has a nice dsl stats page you can go look at.

jadzy wrote:
drop outs in the small hours are likely DLM doing its thing. speeding you up a tiny bit if the line is performing well and can go faster without drops. slowing you down a tiny bit if its not and dropping out. DLM changing the speed causes drop and resync. its part of how your line is managed.

It was this aspect of the original post that interested me...

I have a 38Mbps (ish) FTTC connection (via Zen), to a holiday home which recently started playing up. Remotely, I saw a couple of glitches in its 'Thinkbroadband Quality Monitoring' Stats, then it fell off the internet

Since my home-brew auto-power-recycling gizmo failed to bring it back online, I had to pay the place a visit. When I got there, I found that corrosion had caused the BT Master Socket to disintegrate, which rather explained the problem :lol:

Having failed to take a replacement socket with me :wink: , I soldered the incoming BT wires to a partially dismembered 'ADSL filter' and voila got connected again. My joy was short-lived when I discovered that the download speed was now 1.8Mbs (slower than the upload!). I called Zen, who told me that a 'reset' would probably fix it, but that they weren't going to pass it to Openreach "until it had been stable for 72hrs". If I'd been wanting to use it in anger at this stage, this could have been the cause of a major falling-out! ... but I didn't, so I returned home ... and watched...

The stats. were quite enlightening. After a couple of days of maintaining a connection that would be considered poor for ADSL, let alone VDSL-2, it suddenly took a major leap to 24Mbps ish. After that, it's continued to very slowly, edge back up to what it was - and has recently surpassed it. (There is a reporting delay, between Openreach and Zen of several days)

Code:

Upstream Downstream Sync Date 9997 39956 28-Dec-2020 01:54 9997 38278 24-Dec-2020 00:40 9997 33480 22-Dec-2020 01:35 9997 31010 20-Dec-2020 01:41 9997 28682 18-Dec-2020 00:40 9997 26211 17-Dec-2020 00:43 9989 26211 16-Dec-2020 02:15 9989 23885 15-Dec-2020 13:20 9989 23885 15-Dec-2020 02:34 9542 23885 14-Dec-2020 03:00 2175 1887 12-Dec-2020 01:49 2561 1887 11-Dec-2020 16:38 2561 1887 11-Dec-2020 15:50 2561 1887 11-Dec-2020 15:38 2561 1887 11-Dec-2020 15:17 ::: ::: GOOSED ::: ::: 5354 11056 30-Nov-2020 04:21 6209 16844 29-Nov-2020 19:01 9997 38278 29-Nov-2020 02:20

I just thought someone might be interested :wink:

Sounds like that master socket had become a very high resistance joint! especially if it literally was in bits on the floor Any indication as to why it was getting wet/damp? I'd hazard a guess at water coming in through the wall following the cable. drip loop and better silicone needed? fear you'll get another corrosion problem unless you sort that.

The story and evidence fits fairly nicely for DLM. It would have likely slowed the line significantly having seen increasing errors and lots of drops and resyncs to the point it failed. while failed it would still be looking for a speed (slower) that works. your table doesn't show as many drops as i'd have expected before it died completely. DLM would be trying to find a speed where the line would sync and not just fail repeatedly. you then 'mended' it, which restored it to the point where there was a circuit. so it did sync, but at a horrible speed, best DLM was allowing at that point. DLM saw that and after a few days of being stable and max attainable being much higher that current rate, started pushing the allowed sync speed back up again.

other than the drop on 15-12-20 at 13:20, all the others look the sort of time DLM would be busy doing its thing. some nice passing interference, that hopefully came, and more importantly, went!

doing a DLM reset can be a mixed outcome for ISPs. Likely explaining Zen's position. if your line can only average say 30mbit under DLM management, but you're on a 55mbit service, resetting it might temporarily gets it to say 50mbit, but without any sort of stability. so initially you're happy, you've got 50! but then over the next day or two you notice it drops out lots. DLM slows it over a number of days back to around 30 where its stable. this likely triggers you on the phone to the ISP again complaining about it dropping, complaining about it getting slower etc. so costs the ISP to take your call and manage your expectations etc. maybe attempt to do something about it again, which might not actually help much overall. people understandably want their bb mended and working as well as possible there and then, but its not how the technology manages the EM interference. it uses information over several days to find the 'best' of both worlds.

stats off my router say...
Actual Rate 19512Kbps 2980Kbps
Attainable Rate 24693Kbps 2984Kbps

be lovely if it sync'd at 24. 20 has been the ball pack for months now.

i'm waiting as patiently as possible for FTTP to become available. probably likely many!

jadzy wrote:
Sounds like that master socket had become a very high resistance joint! especially if it literally was in bits on the floor Any indication as to why it was getting wet/damp?

One of the pins in the actual socket was so corroded, it fell off when I touched it. When I removed the wires from the IDC connector, I had to clean them up before I could solder to them - they were very tarnished.

Unfortunately, this is par for the course, as far as this house is concerned - it's a 'rough-built' workers cottage, in Cornwall. Couple the lack of cavity walls, no DPC (and no foundations!) with the county's humidity and you have a recipe for disaster. I've had several mains switches and sockets do the same thing!

that might need a hell of a lot of silicone to fix! ;) or knocking down and building again. maybe replacing some stuff every so often is less grief!

Thought I'd give an update on this. Things got worse early Jan - several dropouts a day. I've tried all modem codes with 3.9.2BT and 3.9.5BT firmware. I bought a Vigor 130 to try just to eliminate any issue with the 2862ac and saw no improvement.
On 10th Jan I replaced the 2862ac power supply with an audio grade 'quiet' power supply from iFi . Since then I've had only one line drop on the 15th and its now been up for 7 days.....
That's enough of a step change for it not to be coincidental in my mind!
I had always thought my power supply was reasonable clean as It goes through an APC UPS, but maybe not. Could still have been Christmas lights related in the neighbourhood of course.
Time will tell!
Steve