July 2009 – Steve Dougherty's Blog

Cables, Infections, and More!

A few days ago I brought my desktop over from Dad’s and set it up at Mom’s, along with a switch to connect both desktops to the network. Someone accidentally bumped the switch, and the cable that brought the network connection up from the basement could not be wiggled back into working again. My hasty crimp had given out. I took my single remaining cable end, and managed to, after maybe two tries, get what seemed like a flawless crimp. Only having one end, and thus one chance at crimping to get online, was a very powerful motivating force. Upon further inspection it is not perfect, but it’s better than other crimps I’ve made. Maybe the loading bars I anticipate getting will help.

I cleaned up a machine at KI that was very badly infected. Our tools took care of the infection, but what took me a really long time to realize was that ndis.sys had been infected and deleted by the scanners. Everything was working except for networking devices, which showed up with corrupted drivers. Fresh drivers did nothing, and uninstalling the drivers did not actually seem to do so. Copying ndis.sys over from another machine fixed the problem.

My Dad sold his house, and we have now moved to Wendy’s house. At this point my room has no blinds, and my desk is in pieces. My room has the feeling of a LAN party in a metaphorical tornado – there are objects one might expect to find in a bedroom, except in strange locations and often not assembled. I hope this gets sorted out somewhat quickly.

AT&T

I feel foolish. I moved the modem to my communications closet so that it would be more difficult for Mom to accidentally turn it off in preparation for a storm, but forgot that I had changed my email password. On my first try I failed to get a DSL signal when running it through a surge protector, so I ended up running it only with an extension. I was confused because the modem’s tests all passed, and both the connection and DSL were listed as up. I could ping whatever I wanted to, although AT&T was redirecting all my traffic to some strange password mismatch page, and traceroutes timed out. (Why not “incorrect?” “Mismatch” confused me into thinking it was set to my mom’s email and my password or something.) I was too worked up over the fact that they were hijacking my traffic to take into account the possibility that AT&T was correct that the password was wrong. I feel bad for wasting their time, but on another note I think that assuring customers that a password utility that requires an executable download is legitimate simply because they continue being redirected to it is not the best way to go.

Zombie 6 Version 2

Zombie 6’s problems seem to be due to a capacitor that was bent until a pin ripped out of it. The only way I might be able to fix this is ordering a new capacitor off the Intertubes, removing the old one, and very carefully soldering on the new one. That likely won’t happen soon, if at all. I got a Pentium 3 @ 866 MHz board with 512 MB RAM from KI. I was very impressed when I was able to just put the board in the case and fire up Debian, which had been booting an Athlon, presumably with an entirely different chipset as well. At this point the ethernet device was named eth1, which was annoying, so I went into /etc/udev/rules.d/70-persistent-net.rules, commented out the eth0 line, changed the name of the new one to eth0, then rebooted. I also commented out the old entry for the optical drive, under which it was on a different IDE channel. For whatever reason, to get networking functional, at each house I had to set the interface to DHCP, restart networking, then set it to static and restart networking again. I’m trying to find out what that does, and why it works.

TX Underrun

Ever since I set it up, once my pfSense box has been running for a while it seems I will plug in the monitor to find two, three, or more messages about TX underruns, such as

dc0: TX underrun -- increasing TX threshold

However, my network started being slow to respond on the 6th, going so far as to drop one packet when I pinged Google. The LAN side was just fine. Yesterday I plugged the monitor into the router to see what was up, as I was unable to find a system log of console messages. The last message was

dc0: TX underrun -- using store and forward mode

My understanding is that this is why it became slow. This machine is not screaming fast by any means. It has 504 MB of RAM and a 499 MHz Pentium 3. Looking in the WAN interface settings, I realized I had misunderstood and set the wrong MTU. I saw the size for PPPoE, and given its mention in my modem’s configuration, I had understood, incorrectly, that my router would use PPPoE. (Even though the WAN interface was visibly set to DHCP on that very page.) I now do not know why I felt the need to set this manually, and as I reread my modem’s status page I feel more and more mystified as to how I came to that understanding: PPP on the modem (Public IP for LAN device). I left the field blank as to allow pfSense to set the MTU, and my WAN connections are speedy again.