Update on my problems, now I've had a bit more time to check it out. It's pretty much confirmed it's a problem between the UBR and my house.
Ping's from my PC to my router or cable modem work great with 100% getting through. Pings to the UBR fail at between 10 - 20% of the time.
Though performance is better than it was a couple of days ago.
I've done a line test with broabandreports and here are the results...
http://www.broadbandreports.com/quality/nil/1493208
Speaking to an ntl person who IM'd me (appologies that I don't know which user that is here
), it would seem that it's a power level problem ("groan... not again" I say to myself
). My upstream is right on the limit, though it has been pretty much since day one of getting the thing. Apparently when it's like this you can get packet loss everytime the modem tries to adjust it's power.
The packet loss is causing intermittent browsing, though the modem doesn't reboot. Basically some requests can stall until refreshed, and this means on most sites I have to use the refresh button a lot or click on links twice to get a response.
I've tried rebooting the modem, the router, PCs and web server, with no effect. I've unscrewed and refixed the coax on the modem and rebooted and no effect.
P.S. A short bit of history...
My upstream power level has always been around the high 50's, often hitting 58, however a while back now I had major problems with the modem failing to lock on the upstream and rebooting frequently. This went on for months, often with days "down" at a time. Calling NTL was no use (40 minute waits in queues, they send engineers and nothing gets fixed, and so on), but Frank and co at nthellworld sorted me out (it had taken a year to fix it effectively by this time). Since then it's been going fine though the upstream still hovers between 55 and 58.
Then the other day I get this reboot, and a new IP address and things start going wrong. I've noticed too that the UBR IP address is totally different (it's now a 10.x.x.x address). Does this mean the UBR has been upgraded and could this mean the whole power level business is up the creek and anyone connected to it may need readjusting again?
mini-rant...
I should point out that I'm never happy about getting engineers out as a) it costs me a lot of money to take a day off work, b) I've had extreme problems before with engineers not turning up (at all, or on time), c) and then there's the hassle of my 2nd phone line installation being screwed up on 4 separate occasions. I've made formal complaints about this and got little other than a months refund and a "sorry, but tough" kind of appology in regards to what it costs to wait in for people who don't turn up and wasted time.
So it's important to understand that I want to be absolutely sure there's a real fault with *my* equipment that needs me to take time off work, and not a problem somewhere else. I know this is a kind of catch 22 as it's difficult to know where the problem is without checking the customer's end first, but you can understand how difficult it is for some of us to keep taking time off to have NTL investigate the problem, and more frustrating when it turns out to be a fault elsewhere.
End of mini-rant there
. I don't want to turn this into a moaning session, and the problems of the past are well in the past now, just something to note that's all, and I'm in no way having a go at anyone here.
Not sure where I'll go with this next, but as no one will be about until next week, I'll just keep an eye on it over the weekend and hope it sorts itself out. It would be interesting to find out if anyone else in the area (GU22, Woking), has a problem though.
update: Upstream power has shifted to 56 now and suddenly I get no packet loss, and things speed up
. Hmm, so it's definitely when it goes over 58 that it happens. I wonder if it will stay at 56, or is this just a late evening dip as people log off?
P.S. other info...
Downstream SNR: 34 dB
Downstream power: 0 dBmV
Upstream power: 56 dBmV (was 58)
Also, when my modem first rebooted a few days back, I was getting a lot of "NO FEC lock" in the log.