Posted by: Ed Tittel
DHCP failure, home network goes on vacation, Windows networking
When I got up on Monday morning, Memorial Day, my wife Dina let me know she couldn’t access the Internet. “No big deal” I thought to myself: “Either I’ll reset the cable modem or the router and all will be well.” Not only was that wrong, wrong, wrong, but unbeknownst to me, I was about to embark on a network troubleshooting adventure of epic proportions. In fact, this adventure is still underway, and will continue later this morning, when Daryl Giles of Austin Advance Technology shows up with his cable testing gear and some replacement parts ranging from in-wall Cat 6 UTP cable to new swap-ins for my punchdown block and network interface patch panel. Sigh.
I did the easy stuff right away — reset the cable modem and the router — and fully expected my problems to be resolved. Not so. “Great!” I figured: “Let’s see what’s up with the network clients.” A quick look at the network status on my primary desktop showed that the network interface was trying to use an APIPA (Automatic Private Internet Protocol Addressing) IPv4 address, which from long experience I know means that the client can’t access a DHCP server. That server runs on the router, so my first thought was that the router had gone south. Jump in the car, drive to Fry’s (thank goodness we have one within 15 miles of my house — where else can you buy an 802.11n/router/firewall box on Memorial Day?), pick up a replacement router, drive back home.
After bringing up a new router, I *still* couldn’t get a DHCP address from the router. My next thought was: “Drat! The cable modem has failed.” Nothing I could do to fix my problem until early the next morning when I could visit my nearby Time Warner service office (just over 3 miles from my house) and swap out my 4-year-old Scientific Atlanta WebSTAR for the now-Cisco-branded 2203C cable modem with phone jack that has replaced the older unit in the interim. After taking the box home, and spending some time on the phone with Time Warner tech support to get everything properly provisioned and working correctly, I was forcibly struck with the realization that some element within my in-home cabling was causing the problem.
Here’s how I finally figured this out: To follow the installation instructions for the router, I had to haul a notebook PC into the master bedroom closet where our wiring center is installed, then cable the PC directly up to the cable modem to launch the router install software (It requires an active connection to the Internet to work properly). I simply ran a 1.5m RJ-45 Cat6 cable from the router to the notebook to do this (you can’t use wireless until the wireless network is setup and configured, of course). This worked fine, after I used the netsh winsock reset command to clear out the detritus of the APIPA setting and restarted the notebook so those changes would take. I happily spent the next 20 minutes or so getting the wired and wireless sides of my home network reconfigured, fully convinced that my troubles were over. No joy!
On the other end of my in-wall wiring, none of the machines attached to the RJ-45 wall plates could access the router to obtain a DHCP address, so they were effectively knocked off the wireless network. I switched some key machines (an HP notebook that Dina is now using for her regular daily computing tasks, and my primary desktop which has an Airlink 101 USB Wireless N 150 network interface filling in for the usual GbE RealTek PCI interface I normally use for network access on that machine) over to wireless so I could get back to work, and called a network consulting company to bring in a cabling technician to check out my home wiring plant.
Right now, I am also using a 100 foot Cat5e cable to hook up my other notebook — my traveling machine, a Dell D620 that’s still running Vista Business because I haven’t gone through the uninstall/reinstall process to move Adobe Premiere from that machine to the newer HP notebook that Dina is using right now. I can’t establish a DHCP connection from scratch using that cable (GbE at 350MHz won’t work beyond 80 feet or so), but because I simply unhooked the 1.5m cable from the Dell while it was in my closet, then attached the hundred footer to its RJ45 port, it can use the DHCP address it already has assigned. Ironically, this machine is 802.11g only, and I don’t want to sacrifice bandwidth by running the router in dual-band mode, so it’s the only machine that currently has full-speed Internet access in the house.
I have to believe that because the punchdown block and network patch panel are the only network components in common among the 4 RJ-45 wallplates in the house, none of which will now resolve DHCP, that the problem has to lie somewhere inside or between those devices. When Daryl gets here later this morning, we’ll figure it out. It’s enough to make a guy like me think more seriously about plopping down the $1K or so it would cost to buy a Fluke Ethernet cable tester. I know it’s got to be an attenuation problem somewhere along the way, but until we can check all the cables and connections (including the patch panel and punchdown block) there’s just no way to figure out what has to be fixed or replaced to get things working again. What I can’t understand is why it just popped up out of nowhere, 4 years after I installed this network which had worked flawlessly until Monday morning, without positing a failure of some kind in one of those components. The cable guy thinks it’s a lightning-induced problem (we had a major thunderstorm on Sunday night) and the networking guy thinks it’s bad cabling. I want to find out for sure what it is, and fix it!
I’ll follow up — hopefully, tomorrow — to report on what we learn and how we fix my situation. Stay tuned!