I didn’t get the shaft this time

I started this morning with a trip to the Toyota dealer — both Priuses needed regular service (oil change) as well as the “60C Special Service Campaign” (aka the steering-shaft recall). I’d called yesterday to make appointments, and the service writer I talked with said to bring in one car this morning and they’d do the oil change and handle the recall; then I could swap cars in the afternoon and they’d do the other one the next day (in the meantime, I’d have one of their cars).

But I had an inspiration — if I was quick, maybe they could do both cars in one day. So I dashed out of the house bright and early and got to the dealer well before 8. I got a different guy as the service writer, and when I told him what I wanted to do, he asked me, “why do you think we can do the recall today?” Apparently there are two different shafts for the Prius, and some cars need one and some need the other, and they didn’t have both in stock.

In fact, they didn’t have any shafts on hand, so it didn’t matter which one I needed. But they did check to see which one, and they claim they’ll order it and notify me when it arrives.

But I still had to get the oil changed, so I waited while they did that — then I went home and repeated the drill with the other car. Fortunately, they have wi-fi, so I was able to work while I waited, but I’d really rather have been able to finish the process instead of having to wait for the shaft.

Stupid KVM tricks

I needed to install Red Hat Enterprise Linux 4 on a machine at work today; normally, that’s a very simple process, marked only by tedium.

Not this time. When I first tried running the install CD-ROM, the screen blanked and I couldn’t wake it up; a friend suggested using the text-mode installer to get around the problem. And, indeed, that did get me through the install — but then when the system booted, the screen went blank and I couldn’t wake the system up.

So I tried reinstalling, with no change; I even tried to use Ubuntu, but it died in the same way.

I thought it must be a hardware problem, but before punting, I asked for help. My colleague came down and walked me through the install, and we had the same results. But then she had a brainstorm — we booted into single-user mode successfully and compared the xorg.conf file on my machine with one that worked. And, sure enough, there were significant differences.

Apparently the KVM on this bank of systems lied to anaconda, so that the installer wrote a configuration file claiming that the display could work at frequencies which were far in excess of what it could really handle. I edited the file to specify horizontal rates around 35 and vertical rates of around 60, and all was well.

Then I changed /etc/inittab to only come up in runlevel 3, which would have worked, too; since I never plan to use the physical console again (it’s in a very loud and very cold machine room), there’s no sense in starting X there.

Moral of the story: don’t panic. And wear earplugs.