Updates

Started: 2003-01-24 12:20:01

Submitted: 2003-01-24 12:50:25

Visibility: World-readable

Hydrogen started turning itself off on an hourly basis Tuesday morning. This was obviously bad, so Scott hunted down an old Pentium II in Craig's garage and we headed down to Inflow. Hydrogen was down when Scott and I got to the cabinet, and the Inflow guy turned it on. I plugged a keyboard and monitor into the console and tried to figure out what had happened. I watched the output of sensors(1) and wondered if the apparently-dying fans were a sensor glitch or the real thing. Scott took the top of the case off and we watched the chip fan spin so slowly that we could see it. Obviously, that was the problem. It was a great relief to know what the problem was, but now we had to fix it.

We concluded the only real way to fix it would be to acquire a new chip fan, preferably one that would blow air sideways instead of out the top, as our current fan did, which was a Bad Idea because the top of the 1U case sat at least a millimeter above the top of the fan.

Since we couldn't exactly acquire a new chip fan on the spot, we turned our attention to the ancient Pentium II Scott brought. I couldn't convince the BIOS to recognize the 80 gig drives, apparently because it was way too old. I attempted to locate a BIOS update, but apparently the only way to get one would be to pay US$40 to some third-party company. Scott called Craig for help and Craig suggested acquiring an IDE controller. We shrugged and headed to Best Buy somewhere on Colorado Boulevard, where we located one for the bargain price of US$50. We acquired the card and headed back to Inflow. Plugging in the card was easy enough, and it loaded lilo without trouble, but it hung when it tried to load the kernel, without bothering to dump any error messages or any such nonsense. I hypothesized that there was something screwy with the controller that prevented lilo from properly booting to it and contemplated solutions.

I booted to the Debian 3.0 "Woody" cd I had on hand and saw that its 2.2 kernel failed to acknowledge the existence of the IDE controller. It recognized the manufacturer id of the card, but not the product id itself, so I hypothesized that 2.4 kernels might know what to do with it. If I had had another Linux box at my disposal, it would be a simple matter to compile an appropriate kernel, throw it on a boot floppy, and repeat the process a handful of times until it worked. Unfortunately, I didn't, and I decided it would be easier to install Debian on the expendable 4 gig drive that came in the ancient Pentium II than to drive back to Boulder, get it to work, and drive back to Inflow. The install itself was easy enough; I've done it enough times that I know exactly what options to pick to make it happen.

While I installed, Scott (and Taylor, while he wasn't doing something or another with the remainder of the cabinet) tried to redirect the airflow in Hydrogen to make it cooler. I was afraid that it was going to look like it would work now and then explode horribly sometime late in the night. I managed to download 2.4.20 onto the new 4 gig drive and compiled it with options that looked right. When I rebooted, the kernel saw the controller and even went to the trouble of auto-detecting and setting up software RAID. I instructed lilo to boot to Hydrogen's mirrored drives by default, rebooted, held my breath... and it worked.

At that point, I was about ready to elevate myself to godhood, for at least the third time.

(I don't specifically remember previous times when I elevated myself to godhood, but I'm sure there must have been at least two of them.)

With our mission accomplished, Scott and I put Hydrogen(II) back into the cabinet itself and departed. Scott took Hydrogen's black 1U case home in an attempt to hunt down a suitable chip fan. Hydrogen(II) seems to be working rather well now, which I can definitely approve of.

Tuesday was a fun-filled and action-packed day for me; in the morning, I headed over to iTi, conveiently located in Boulder, for something closely-resembling a pre-interview. They can't actually hire anyone until they get the contract they're working on getting; they should know in February sometime. The tiny company makes industrial ink jet printers and a handful of other stuff; it looks like fun stuff. I like the idea of working there; it's a small company (I think I shook hands with almost everyone), so I will get to do a lot more stuff than I would at any large company.

(Which reminds me, I need to call a certian larger company, since they said they'd call me by the beginning of this week, and it's most definately not the beginning of this week any more.)

I imagine there are more things I could talk about, but this sounds like a good synopsis for now.