Hosting Problems

If you’re one of my eight or nine regular readers, you may have been wondering just what the heck has been going on wiht my blog. The rest of you can ignore this…

In the space of 7 days, my blog server crashed 22 times. I finally realized that it was crashing every time it did certain things — and all of those things involved a particular set of hard disk drives (a mirrored pair). Once that dawned on me, I had a way to troubleshoot the problem, which I managed to do yesterday — though in the process I managed to crash the server so badly that it wouldn’t start up again.

So…it was off to Fry’s again, for two more hard disk drives (amazingly cheap: 160GB Seagate drives for $89!). Then off to my hosting site to grab the (dead) server, back home to pop the new disks in and perform all the nerdly ceremonies required to bring the server back from the dead. Finally, at 9 pm last night, I was able to run the server back to my hosting site and pop it back in — and it’s been working just fine for the succeeding 10 hours.

At this point I’ve replaced the power supply and five disk drives in that server — and it only has four disk drives in it (one of them was replaced twice). All of these replacements are within the last 30 days, and this is a server that’s been working without even the hint of a problem for the preceding 3.5 years.

I believe I could be forgiven for thinking my hosting site was haunted by grumpy gnomes who spend their days inside my server, hacking away with miniature sledgehammers until they make something break.

But I suspect the cascade of problems is of more prosaic origin: most likely four of the disk failures are the direct result of the earlier power supply problem. If I were to guess, I’d say that the +12V power (which is used by the disk drives, but precious little else) got way out of whack on that bad supply, and basically “cooked” the four disk drives that were in the server at the time. There was one earlier disk drive failure that might have been independent, or it might have been an early casualty of the bad power supply (which appears to have degraded over a period of weeks).

But now the power supply and all the disk drives that it ever powered have been replaced. If the server is now reliably “up", I will assume that evil power supply was the root of all my recent problems. But if the server starts flaking out again, I’m heading over to my hosting site with a big can of Gnome-B-Gone (guaranteed to kill all gnomes in a hideously slow and painful manner!)…