Server Troubles
Had a major scare last night which thankfully turned into a minor one. It initially seemed as if I had a hard drive failure on Glucose, my webserver, which is just a glorified desktop in a micro-ATX case, complete with patched-up micro-ATX power supply. The power supply’s fan failed some time last year, so I attached an auxiliary fan to its exterior shell, removed the dead fan from the interior of the power supply, and it has pretty much held ever since. Now it seems that the supply is dead or dying (possibly due to the weather), as the whole system shut down yesterday and failed to restart, or it would start, then quickly die in an endless loop. So the hard drive was full of errors, and I’d fsck it, then it would crash again. Naturally, my first thought was that it was the hard drive.
Yes, my not-backed-up hard drive. R asked if I had implemented a proper backup plan. I had not. Rather, I backed up everything, including the database … but everything went on the same drive — hence, the improper backup plan. I shuddered a little when I thought of losing 6 years of blog entries.
Well, it wasn’t the drive this time. I got lucky. The drive is now sitting in an antiquated PIII/600 box which is housed in an impressive generic rack-mount case, which looks like it may be the new home of zuckervati.com, if I can build some redundancies into it, such as a RAID1 (at least). I even had some notions about dropping the whole thing into the enormous and even more antiquated NetFinity server, which seems to be built for this kind of thing.
But I’m not ready for that. Firstly, it would require I purchase several decently sized hot-swappable SCSI drives, which I can’t afford right now. Either that, or configure the array of 9 and 18GB drives into some kind of RAID 1+0 array, which might give me a nice redundant 27GB, but I’m not even sure those drives work. I’ve never even booted that beast, afraid of a large scale power outage in my neighbourhood.
I was thinking of donating it to those guys at OpenBSD, since they look like they could use the hardware — and the OpenBSD system has served me so well these last few years (this was clvrmnky’s idea). It’s quite resilient, and transplanting a server drive from one machine to another like I’ve done is proof enough of that.
Expect the occasional outage this week, while I get things stabilized. First thing I’m going to do is backup everything.