> At my last place, I ended up writing a small Tcl/Tk script to do more or
> less the same thing (but with trippy statistics like standard
> deviation). I also wrote a script to do trace-route against half a dozen
> Internet hosts so if the link went down, I could tell where it went
> down. (Half the problem with spotting fault behaviour is knowing what
> normal behaviour looks like.)

Well on my modem "normal" behaviour is a ping of about 35 ms to google 
and 5 packets received out of 5 sent. Above 100 ms or less than 5 
received is not normal.

So just to update, after 3 calls to India an actual physical real person 
came round to my house to have a look. He tested the line from the 
socket and also the outside of the house and said actually the line was 
better than average. He replaced the wall socket and gave me a new modem 
anyway to rule those out, but couldn't find any problem at all with the 
line. Interestingly initially the modem was connected at 5mbit instead 
of the normal 3, and ping was down to under 20ms to google. But after a 
few hours the ping suddenly jumped back up to 35ms, and the speed has 
gradually come down to about 4.5mbit now.

He had a chart on his laptop that showed the connection quality for the 
last 14 days, and the red blocks matched my chart :-) But he had no idea 
why it was behaving like that :-( At the moment I can't be bothered to 
take another 1/2 day off work just to be told the same again, but I'll 
continue to keep logging...

I suppose ideally he would leave his DSL analyser box thingy connected 
to the line for a few days until it drops, but BT probably don't do that 
for a mere residential customer.

Post a reply to this message