Anybody else seeing outages on RoadRunner in the early morning hours? I've noticed it for several weeks - being six hours ahead in England, 2:00 - 5:00 AM here is 8:00 - 11:00 there, prime time to connect.
I'm wondering if it's something in the nightly maintenance run on this old PC that runs the firewall. Never used to bother it that I know of, but the timing's right. It's running RH 7.3.
I need to find a way of snapshotting the load on it in the wee hours when I don't want to be up.
On Thursday 26 May 2005 09:15, Jonathan Hutchins wrote:
Anybody else seeing outages on RoadRunner in the early morning hours? I've noticed it for several weeks - being six hours ahead in England, 2:00 - 5:00 AM here is 8:00 - 11:00 there, prime time to connect.
I have noticed it too. I was having connectivity issues 5-6 times / day where the connection would drop offline for about 2 - 3 minutes and then suddenly return -- on rare occasions, for an hour or two. It turns out that the copper braided shielding on the coax line leading to my house had a burr in it. More on that in a minute. I collected some ping logs and used gnuplot to make a PNG that someone at RR would be able to easily interpret. The short version is that, after 3 techs on-site and several phone techs whom all but called me a liar, they finally fixed that but this problem you describe still exists. I'm thinking of starting the customer service cycle over again for this problem but am dreading the amount of effort it might take to convince them.
The trouble is that it's not consistent. I've noticed it as early as 3:30 AM and as late as 4:35 AM. And it's not on every night.
Anyways, I'm going to start making ping logs again. ("ping <next node in network > pinglog.log") The trick this time is going to be to ping the next hop in the RR network and something guaranteed to be outside their network simultaneously so that I can show without a doubt that it's a node in the RR network to blame.
I'll post the gnuplot script here when I write it.
On that burr problem... it's really interesting to me that this was happening:
(ASCII diagram follows)
+-------+ <------- outer black plastic |+-----+| <----- copper braided shielding ||+---+|| <--- white plastic shielding ||| * ||| <- copper data core ||+---+|| |+-----+| +-------+
So, the troubled line was installed after the last ice storm in near-dark conditions very quickly by a RR sub-contractor as they were trying to restore service as fast as possible. What happened is that a piece of the outer copper shielding was frayed off and became wrapped around the inner copper data core. So, the exterior shielding was acting as an antenna and the signals it picked up were carried on to the data core. Unfortunately for me, some EM interference source from one of my neighbors was close enough to the RETURN data frequencies used by RR to interrupt my service for however long it may be. I'm guessing that it was a cell phone due to the nature of interruptions. The weirdest part of all of this was being able to use Ethereal to see ARP and TCP traffic _inbound_ but being completely unable to send anything out.
Anyways, I found it interesting.
You might find "mtr" more useful than ping - it'll show where in the route the packets are being lost. Used that this week to determine that it was at a nuvox node near epic aliance's cave. Not that we could do anything about it...