(no subject)
Mar. 22nd, 2007 06:49 pmJust got this email from the University of South Carolina's engineering department:
"The network problems that we had Tuesday night have replayed themselves 90-fold this afternoon starting at about 1:30pm. There was again extreme slowness on the College network causing an outage. It was traced to a network bridging loop in a different research lab this time. We disabled the connection to the lab and rebooted the router to restore the network to working order. The network was up and down while we were tracing the problem down. It has been up for about 1.5 hours now, and appears stable.
Again, please be extremely careful when using mini-switches in your offices and labs. One of these devices improperly connected can bring down our entire network. If you are not sure how to set up a mini-switch, contact your departmental IT administrator for help."
Oy vey. I shudder to think what kind of loops these professors put into their research.
"The network problems that we had Tuesday night have replayed themselves 90-fold this afternoon starting at about 1:30pm. There was again extreme slowness on the College network causing an outage. It was traced to a network bridging loop in a different research lab this time. We disabled the connection to the lab and rebooted the router to restore the network to working order. The network was up and down while we were tracing the problem down. It has been up for about 1.5 hours now, and appears stable.
Again, please be extremely careful when using mini-switches in your offices and labs. One of these devices improperly connected can bring down our entire network. If you are not sure how to set up a mini-switch, contact your departmental IT administrator for help."
Oy vey. I shudder to think what kind of loops these professors put into their research.
no subject
Date: 2007-03-22 11:13 pm (UTC)I'm assuming a script which momentarily chops the noisier segment of the network into smaller and smaller binary tree halves until it can't narrow it down any further wouldn't be of assistance?
(Bonus points if it narrows down the exact port(s) causing the problem and disables them until manually reinstated.)
no subject
Date: 2007-03-25 12:53 am (UTC)no subject
Date: 2007-03-22 11:26 pm (UTC)no subject
Date: 2007-03-22 11:27 pm (UTC)How does this happen, exactly? I've tried bridging two ports on a switch, but nothing interesting happened. Do you have to bridge two switches together through another switch to summon the etherspooge?
Yes, I really am that easily amused.
no subject
Date: 2007-03-23 12:15 am (UTC)Had a luser that brought in a LinkSys ROUTER and plugged it in backwards. On a Monday morning. For about two hours
Took me a while to figure out why the internal DNS was all fraked up.
no subject
Date: 2007-03-23 02:09 am (UTC)They're frighteningly quick to respond with a dhcpoffer, beating out most real software-based dhcp systems with thrilling speed... then doing really fucked up things once they exceed their pool of leases (configurable on things like dd-wrt, but fixed to 25 or 50 on most).
no subject
Date: 2007-03-23 02:48 am (UTC)no subject
Date: 2007-03-23 12:09 am (UTC)We now have filtering enabled that shuts down most ports as soon as it sees more than one concurrent MAC address in packets travelling over a given port. I forget the term the network guys used for this filter...
no subject
Date: 2007-03-23 12:40 pm (UTC)Your network must have been fuxx0red to be listening to BGP routes from routers it doesn't know about. You can filter those packets out on all the ports _except_ the ones you know should have BGP traffic on them. That's just asking to b p0wned
no subject
Date: 2007-03-24 02:06 pm (UTC)yeah, the network's a ton more idiot-proof than it used to be, but it's not that easy sometimes to idiot-proof the original idiot-proofing, I guess...
no subject
Date: 2007-03-23 12:35 am (UTC)no subject
Date: 2007-03-23 03:12 pm (UTC)Also. I feel a carefully measured proportion of your pain.
no subject
Date: 2007-03-23 06:11 pm (UTC)The data centre I used to work for configured individual VLANs for all their users to prevent just this kind of thing from happening; they forgot to implement this on one switch, however, and said switch was brought to its knees by a broadcast storm when some dumbass disabled STP on a switch and plugged it in.
no subject
Date: 2007-03-25 03:36 am (UTC)He brought down his subnet. Three. Times. In one month. After I repeately told him that personal wireless access points or routers were prohibited. After I repeatedly told him that he was hooking up his base station incorrectly and it was trying to assign IP addresses to the other people in his subnet.
I finally told him that if he did it again, he was guaranteed to have his network connection terminated for the rest of the academic year. I think that finally made him quit.
However, this is why the residence halls/apartments are on an entirely separate network from the rest of campus - and split up from there. It used to be that some nob freshman in the dorm would kill half the campus network...