BRAS drop on a stable line

My line had a couple of resets around 2 and a half days ago (due to a storm) and ended up with a line rate of 5696 Kbps, slightly below "normal" but not far away. I've been doing one of the new BT line tests around 07:00 every morning and the last 3 days shows (profile / rate / throughput)

17th 4000 / 5696 / 3672
18th 4000 / 5696 / 3662
19th 2500 / 5696 / 2290

Line up time is currently 2 days 11 hours.

I knew that a BRAS profile will change upwards if the line is stable but I didn't know, and don't quite see why, it can also drop for a stable line. Any comments?

are you certain it hasn't sync low recently ? If it hasn't then it looks odd - almost the wrong way round, taking 3 days to respond to the thunderstorm.

In reply to:

---

are you certain it hasn't sync low recently

---

Definite. I have a process running that logs selected router stats every five minutes and that shows the last resync took place on the 16th at 21:50. All the other readings, error counts etc, agree with that being the last resync.

I've just done another speedtest and that agrees with the last one - a BRAS of 2500, sync of 5696 and throughout of 2291.

I'm glad I'm not the only one who finds it odd

I've just had another resync. SNR was just above 9 at the time so I don't think it was as a response to a line problem?

I've now got a BRAS profile of 2500, a line sync of 6688 and a throughput of 2301. My line can't normally sustain a sync of over 6100 so I expect I'll have another resymc later this afternoon/evening. I hate to think what _that_ will do to the BRAS profile.

the resync may be due to errors.

In reply to:

---

the resync may be due to errors.

---

In the half hour before the resync (Down)

SF Errors 2
RS Uncorrectable 21
ES 2

Up errors were all 0

<Shrug>

if 10,000 errors came down the line and caused a resync you may not see them if the counters zero out.

True, but some of the counters don't reset. Errored Seconds, for example, went from 2495 to 2505 - and that includes any errored seconds caused by the resync. I suppose it is possible that there was a massive burst of noise for a second or so and that triggered the resync but I would have thought there would be some other indications [1].

BTW, I'm not just being argumentative here - I do appreciate the suggestions.

[1] I have got a local source of interference that turns on and off, shown by a very sharp +- 3db change in my SNR and usually lasting for an hour or so. It comes, I assume, from a local small industrial estale with which my line shares a 50m stretch of aerial cable. It's the main reason why my line won't sync at 6800 for any length of time.

it was the thousand errors in a second scenario that I was guessing at for causing the resync. Either that or an instant SNR reduction.