Weather makes no difference. All the cabling is under ground and the line is better at night when it's colder (i.e. the 5pm thing isn't a temperature issue).

Edit: Furthermore, the rain can't arrive at 5pm every night. Not in Scotland.

Edited by camieabz (Thu 04-Oct-12 22:07:18)

Nah ! BQM needs to be a visible link. I've just re-jigged my sig to include it.

(Also @WWWombat) As for the sync on the graph, I never could get my head round the full RouterStats. Lite is better for my brain.

"All Connection Data"

Two more characters in my sig and it rejects it .

In reply to a post by camieabz:

Weather makes no difference. All the cabling is under ground and the line is better at night when it's colder (i.e. the 5pm thing isn't a temperature issue).

Edit: Furthermore, the rain can't arrive at 5pm every night. Not in Scotland.

I agree that the 5PM thing seems more likely to be crosstalk. But that doesn't mean the weather does *nothing*. One extra bit of rain, plus one extra crosstalking modem might make the difference.

Well I can't do anything about the rain, so let's assume it's not the rain.

Just to update, it popped back up to normal SNR at 23:45 ish.

Rebooted and popped back to 15,500, and now the SNR is dropping...

I was on gateway 01, and now on 08. I wonder if Plusnet's gateways 04 and 08 issues (per status announcement) are causing this latency.

I wouldn't be surprised at slightly different latency, as on your BQM after the resync, on different gateways. But that is nothing to do with the SNRM.

You didn't give us the Routerstats graph following the one you supplied. One of the major advantages of RS over DMT is that it continues after a resync. The behaviour then can give a clue as to the type of event that caused the disconnection.

It looked like you have a fairly high sampling interval as well. That's fine for a rough guide, but for serious diagnostics I much prefer 5 seconds. Logging to file as well as graphing can be useful as well, if you aren't doing that. Scanning the log is sometimes easier than the graphs if the person looking isn't using the machine the graphs are on, so can't flick through them.