Print

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

[jelv]

I presume Weaver's lines don't have telephone numbers associated with them that he could ring - shame as that would definitely be worth a try.

[burakkucat]

The lines do have telephone numbers allocated to them but they do not carry telephony services.

[Weaver]

Burakkucat has all the phone numbers. I agree with his thinking on this.

Update: Today Janet plugged line #2 in, the first one to fail, after 9 days unplugged, and - tara[!] - it’s suddenly 100% perfect! I don’t know what has been done to it; maybe AA or someone effectively phoned the line, or initiated a disruptive test sequence. I did mention the wetting ring-ring theory to AA and told them it was only a hypothesis, but maybe they tried it.

So now we still have line #4 ill, at <50% of usual downstream sync - only 1.4 Mbps d/s sync instead of 2.9 Mbps. So that needs fixing by magic remotely or by an engineer visit. I sent AA the visual proof of the illness - an image of the SNR-vs-tones of the link from the Johnson custom firmware ZyXEL VMG 1312-B10A modem’s easy-stats service (as I have termed this Johnson server), image with a horrible shape and a deep hollow where the max downstream bit loading should be, in the tones 40-90; 11 bits per bin down to 4/5.

[jelv]

I did mention the wetting ring-ring theory to AA and told them it was only a hypothesis, but maybe they tried it.

If that's something they could do, you should be able to do it also. What happens when you dial one of the numbers? If it comes up number unobtainable I don't believe they could do it.

[burakkucat]

What happens when you dial one of the numbers? If it comes up number unobtainable I don't believe they could do it.

Please review all that has previously occurred in this thread . . . In particular Reply #8 and Reply #9. 

[Weaver]

Engineer came on Wed afternoon and rang Janet ahead to tell her he was one hour’s drive away. I looked at the modem and it was perfect!

ayes, same story again.  Mystery fix, no clue how.

Engineer arrived, Janet and I told him the story, he tested everything as usual and told me he had superb figures for balance etc. We had a chat afterwards and I showed him the old SNR-vs-tones graph, as opposed to the now perfect current picture. He didn’t do anything notable before he came. I asked him about water ingress but we agreed that it had been raining heavily for weeks, so our mutually agreed conclusion was that that day was nothing special. I suggested to him that AA might have run some kind of remote test earlier in the day. Ahead later moaned to Janet about ISPs doing things that they ought not to, not that Janet was to blame for anything. 

Later on I looked in the clueless.aa.net.uk log for the line; there were no relevant events that afternoon.

Emailed AA to tell them the story. engineer sent his notes in, the usual ‘no fault found’ which was quite correct.

So we’re now more bemused than ever. Could it be that unplugging a line from a modem might change things ?

Could it be a modem fault? Mind you, we would have to have two faulty modems, which seems incredibly unlikely, as there were two faults in the same week. Could a now-you-see-it-now-you-don’t modem h/w fault affect only tones 40-90 ?

For some reason, Janet talked me into swapping two of the modems: those labelled #3 and #4. This was a bad idea but I was very fuzzy, being bombarded, and agreed to something that I shouldn’t. When my mind cleared, I made Janet put the modems back the way they were. Now anyway, this involved possibly removing power from the modems, certainly unplugging each from its cables to the MUX/DEMUX VLAN switch that they’re connected to (between modems and the router).

So:
1. the line was disconnected from the modem (twice) - so a possible effect on the line, and
2. (possibly) the modem was rebooted (twice); in fact I’m pretty sure I remember seeing evidence that it did get rebooted because some of my network state overview programs moaned about not being able to access a modem.

So it just might be that some of those actions affected either the line or modem in some beneficial way. I can rule the modem out by swapping it for a new one, but I’m not remotely expecting to find anything as it just doesn’t feel like that kind of bug.

Perhaps we should go cargo-cult: have a fake engineer visit to fix the next fault - get a friend to drop in pretending to be an OR engineer and save AA some money.

[burakkucat]

I am unable to think of any rational train of events that can account for the problems you are, quite clearly, experiencing at random times. 

[Weaver]

I’m wondering if the ‘cure’ is even correlated with any events happening around the time of the engineer coming out, or earlier, or while he/she is here. Is it possible that the problem is in some sense ‘fixed’ at an earlier time but I don’t know about it until some action is performed? - such as unplugging the line from the modem, or forcing a retrain, or even forcing a reboot. That is, the problem actually becomes ‘fixed’ earlier but I have no way of knowing about it/realising it until someone happens to do a certain something later.

[burakkucat]

Yes, that is a possibility. But how can one test that scenario? 

The next time you notice the problem, perhaps --

• Disconnect the modem from the NTE5.
• Power down the modem.
• Wait 24 hours.
• Connect the modem to the NTE5.
• Power on the modem.

(Notice that I have deliberately suggested performing the first two steps the "wrong way round".)

[Weaver]

Will definitely do exactly that. Thank you.

[Weaver]

And guess what. On Thursday, the line #4 went back to its bad old ways but far far worse than ever before. A magnificent 242 kbps downstream. I now claim the title of slowest line of any kitizen! Can this be officially confirmed now please

Have informed AA. Next I will follow Burrakucat’s procedure above.

[burakkucat]

A magnificent 242 kbps downstream. I now claim the title of slowest line of any kitizen! Can this be officially confirmed now please

Certainly. 

I have just had a sudden memory of using a remote full-screen editor, at 300 bps, way back in the 1980s. 

[Weaver]

[ Dear admins, could we perhaps split off this recent slice of intense nostalgia for the horrible speeds of the Pre-Cambrian Era? ]

[Moderator note: Done. See Nostalgia - Networking in the "Early Days".]


Have now completed the Burakkucat Procedure. Did indeed:

• Disconnect the modem from the NTE5.
  
• Power down the modem.
  
• Wait 24 hours.
  
• Connect the modem to the NTE5.
  
• Power on the modem.

So modem is now back on. And once again it’s now perfect!!   

Sync: 3027 kbps down / 399 kbps up
Attainable: 3396 kbps / 328 kbps
SNR: 3.3 dB / 5.8 dB
Attenuation: 64.0 dB / 40.3 dB
CRC: 2 / 0
ES: 1 / 0
SES: 0 / 0
FEC: 161 / 38
Link Uptime: 2 hours, 3 minutes, 16 seconds

So many thanks to Burakkucat, whose instinct proved once again correct.

[Weaver]

Discoveries:

1. I’ve found that a reboot does not clear the problem. ie using the modem’s command line reboot command via telnet.
   
2. A power-cycle does cure the problem! This is physically pulling out the dc for a very short time. I have not tried the modem’s on/off switch. Is that a simple mechanical switch? - as opposed to a reboot interrupt button which causes a controlled shutdown.

Janet doesn’t think the OR engineers turned off my modem. Indeed, she has a point: why would they.

So we ought to also test the following: 

3: test whether or not just pulling out the phone line alone fixes the problem. 
4: swap out the modem for one that has never been in use.

This way we can find out whether it is a modem gone bad or weirdness in the phone line. I think that when the big lightning strike happened in jan-feb 2020 (I forget the date, would have to look back) it is possible that all the modems got mildly ‘cooked’ so that even though they appear ok, there is actually something wrong with them, and this something causes the hollow curve phenomenon every once in a while. So will perform those tests as soon as I can procure the services of my beloved wife, who is knee-deep in govt paperwork.

So test 4 is important; it has to be a new modem, one that was not affected by the great lightning strike. I can’t just use modem #3 which was in use from before jan 2020 until summer 2021, and so was exposed to the strike.

So anyway, more tests to do but at least we have a fix now, one that we ourselves are in control of.

[tubaman]

If a power cycle fixes the issue but a soft reboot does not then that suggests the soft reboot does not fully clear all memory etc.
I doubt it's a hardware fault but there's no harm in trying a different modem to be sure.
The power switch on the VMG1312-B10A is a mechanical switch that disconnects the 12V power so will perform the same function as removing the power adaptor from its 13A socket.