Print

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

[Weaver]

Bad thunderstorms on Saturday, people killed and injured on the mainland. Took all lines down as a precaution when it came near, as I had been watching its approach for several hours. Lightningmaps / Blitzortung app warned me sixty miles away and my hardware lightning detector flashed about far distant strikes, so both did a very good job with plenty if warning. The alarm been on the lightning detector is almost inaudible unless your ear is right next to it which is ridiculous, but I saw its bright blue led flashing for a strike together with other colours, so all was well.

When I took all the lines down by disconnecting modems from the wall sockets the Firebrick instantly switched over to 3G (I don’t have a 4G NIC for some stupid reason). Same IP addresses maintained, TCP connections unbroken so no disruption to transfers in progress! Good job Firebrick people. 

About an hour and a half after that some packet loss was visible in clueless for line 4. Continued all night and all next day with several retrains. Downstream sync rate has dropped from ~2700 to ~800kbps. Upstream rate is unchanged.

Have emailed AA.

[burakkucat]

When I took all the lines down by disconnecting modems from the wall sockets the Firebrick instantly switched over to 3G (I don’t have a 4G NIC for some stupid reason). Same IP addresses maintained, TCP connections unbroken so no disruption to transfers in progress! Good job Firebrick people. 

About an hour and a half after that some packet loss was visible in clueless for line 4. Continued all night and all next day with several retrains. Downstream sync rate has dropped from ~2700 to ~800kbps. Upstream rate is unchanged.

My understanding is that you reverted back to the usage of the four metallic circuits, post atmospheric electrical disturbances, within an elapsed time-frame not exceeding 90 minutes.

Either a near-by ground strike has caused detrimental changes to just one circuit or the haggis are still performing acts of revenge, whilst no one is about to witness their attack.

[Weaver]

In the end it seems to have been resolved without any corrective action.

There were strikes on the Sléite peninsula fairly close to us, to the south of us, but our line runs northwards to NSBFD.

Whatever it was, by early the early hours of Monday morning the packet loss had gone away, so what it was all about remains a mystery. I put the downstream SNRM up to 12dB when I noticed the problem just to overcome the packet loss. Then later on when things looked ok, I put it back to 6dB then 3dB. Speed is good now.

[Weaver]

If line 4 was slightly baked, by induced current, as Burrakucat suggested to me then why only line 4? Given that they all run adjacent to one another. Weird.

[burakkucat]

Weird and unexplainable. 

[Chrysalis]

Weaver did you still not look into 4g? as the issues keep happening with the unreliable ADSL.

Line 4 could be at the edge of the bundle so weather triggered problems could affect only that line if the other lines are not in that part of the bundle.  These are twisted pairs and are in a bundle with lots of other cables.

[Weaver]

Hi Chrys I’m afraid that I have been so ill recently that I haven’t had the energy to get into it.

> as the issues keep happening with the unreliable ADSL

Actually it’s the reverse. The current system works so incredibly well, because of AA+Firebrick. A line goes bad and no one notices it. It would be totally different if it was a single line setup. What happened this time was that my tweet-from-AA facility is broken (at AA) and because my AA-SMS-wife-me path is broken, at the wife node, then I never knew for 24 hours and my wife didn’t complain about bad internet. So it was only when I happened to take a look at AA’s clueless.aa.net.uk server and saw red all over the place on line 4 that I knew I had a problem.
AA’s Firebricks (or whatever) had done a switch-over of traffic fraction to the good lines.

I am thinking about what you said about ‘the edge of the bundle’ - if it isn’t that then what else could it be.

[burakkucat]

Do we ever hear about exchange based equipment being zapped by atmospheric electrical disturbances? Virtually never. (I type virtually as I am sure Black Sheep, licorice or 4c would be able to mention one such an extremely rare occurrence.) There are serious protection devices attached to each pair which, coupled with a very low impedance connection to ground, can cope with most induced high-voltage surges. The gubbins can display suicidal tendencies, if really provoked, as a means of last resort to protect the exchange equipment.

So I postulate that a nearby ground-strike was enough to induce a sufficiently high voltage in the pair of your line number four. The gubbins, at the Broadford exchange, did its best and, as a result, received a terminal blow. The Broadford exchange, just like many others, is not manned (or womanned) by its own resident engineer so any alarm would be passed to some operational centre for attention. End result being that Hamish (or possibly Duncan) attended and replaced all the protective gubbins so affected, for your line number four and for others locations in Heasta, returning all to normal service.

But, as I typed at the beginning of the previous paragraph, that is just my postulation which fits what you have observed and reported.

[Weaver]

That is a good account.

So does chrys’ outside-of-bundle idea account for that pair being different? and then that can link into your story.

Could such protective equipment receive a blow and still leave the pair operational for dsl albeit impaired ? Because it was still working, just at 25% of the downstream sync rate.

This was at the weekend and in the early hours of Monday am I increased the d/s SNRM to 12dB or something, enough to stop the packet loss. I saw no reason to take the link out of the set, unless a line that slow might be causing trouble, because it was still providing normal very valuable upstream.

Here is the graph for last week, starting on the Saturday when I took the lines down.

[burakkucat]

So does chrys’ outside-of-bundle idea account for that pair being different? and then that can link into your story.

Yes, indeed. I would go as far as to say that it is an integral component of my postulation.

Could such protective equipment receive a blow and still leave the pair operational for dsl albeit impaired ? Because it was still working, just at 25% of the downstream sync rate.

Gas-discharge tubes, heat-protection coils, silicon-carbide disks, fusible links, who knows what. If one leg of the pair remains connected to the exchange equipment whilst the other leg has been disconnected (by one the the aforementioned protection devices) an xDSL service could still limp-along. Exactly as you have described. 

[Weaver]

Thanks good to know!

[Chrysalis]

lines on the outside of the bundle logically would be more vulnerable to environmental effects but less prone to crosstalk, those in the centre would be the opposite.

[Weaver]

The other bewildering difference between my pairs in the same bundle is the up-down jumping squarewave of the upstream SNRM. Line 3 is the famous period=1 day 4dB-high example, and line 2 wiggles up and down by 1 dB with a less predictable waveform of period which is - very roughly - a few hours - maybe 5/6 hrs.