Into The Weeds

So at some point this afternoon, this server took it’s well-worn uptime of 250+ days into the weeds. I was at work at the time, so could only speculate as to what may have happened. It would take a prolonged power outage to knock it completely offline, since it’s protected by UPS, and that was definitely not what happened — the router is plugged into the same UPS, and it had not gone down. My next guess was that the heat had gotten to be too much (it was about 101 degrees here today) and the AC had failed and there was some sort of fan failure and/or thermal shutdown.

I started the troubleshooting when I got home around 5pm, and immediately ruled out thermal shutdown since the fans were still all spinning and things didn’t seem too warm in there. It had been warmer than usual in that room, so it’s possible that the heat had taken out some other random component. Plugging in a monitor and keyboard and rebooting, I could see that the system would engage the monitor, and then beep wildly and/or fill the screen with garbage characters. Unfortunately, it’s been a while since I’ve had to do any PC hardware troubleshooting, so I started barking up the wrong tree by replacing the processor, and then the entire motherboard and video card (I have spares of each for the server). Now, with most everything unplugged, and the new motherboard in there, I could get it to at least POST and get into the BIOS, but as soon as it started to initialize the SATA card in there, it would hang and the drives would make funny noises. Can you guess what the problem was? So as not to spoil the “surprise” and how the rest of my evening went, I’ll continue this in section 2.

I went ahead and totally unplugged everything, including the two EIDE drives in there and the ethernet card, and the SCSI card for the external drives. On the next power-on, the SATA drives (which I boot off of) spun up and went into boot sequence. At this point I finally realized (about three hours too late, in my opinion) that it was a power supply problem. Of course, the last time I had a crash it was a power supply problem as well, and likewise took me way too long to figure out.

Luckily it was just after 8pm, so MicroCenter was still open, and I headed over there. I always liked MicroCenter — it was the place you could go for a power supply, or a motherboard, or random computer components, and they had good selection too. I was afraid that in the 5 years or so since my last trip to MicroCenter that they had become all “Best Buy” and would only have one power supply that was neon and cost $200. Much to my relief, it was the same old store. They had a selection of about 15 power supplies ranging from the cheap $49.99 to the fancy $200 750w+ with neon. I settled for something in the middle and was on my way.

On my way home, I ended up stopping at the Whole Foods market nearby on River Street and picked up a very expensive, but amazing looking piece of N.Y. strip steak to reward myself for getting this taken care of. I grilled it up, and man it was the best store-bought piece of beef that I have ever had.

So finally, the server is up and running, but still on the “backup” motherboard which has a busted EIDE controller so local disk-to-disk backups aren’t working. Tomorrow I’ll probably get around to swapping that out again, but all in all things didn’t turn out so bad.