Around 2 PM today, I got a blue screen on “vector”, my web/media/SSH/everything-else server. That was unusual. I haven’t seen a blue screen in months, probably not since I last had problems with my main workstation [1] in January.
I rebooted and noticed Apache wasn’t running. I tried to manually start the service, but it “terminated abruptly”. I read the errors logs. I ran Apache from the command line. Nothing. I looked online for help on running Apache on Windows and found an explanation that kept referring to an executable named apache, which I didn’t have (mine was named httpd). In general, the document made Apache 2.2.4 seem a lot smoother than Apache 2.2.2, which I had been running, so I decided to uninstall Apache 2.2.2 and then install 2.2.4. Then I’d merge the httpd.conf files. That should fix Apache not running, right?
Well, the install went smoothly and Apache happily displayed it’s “It works!” webpage. Then I merged my changes into the httpd.conf files, and suddenly Apache wasn’t starting anymore. I undid my httpd.conf changes, and it was starting again. I looked through the changes, and they all looked innocuous: A few default permissions, the list of files that should be treated as text, and the document root.
The document root! My D:\ (VECTOR_MISC) drive had disappeared! I rebooted several times, tried an auto-detect in the BIOS, and it still hasn’t come back. I haven’t actually opened the case yet to examine the hard drive (maybe I should have done that first), I wanted to restore a working computer first, which meant changing drive letters and the document root.
The drive that has apparently died contained my web server’s document and nothing else. I also have a DVD backup from January 24. The local copy on the MacBook might even be a bit more recent. Probably not that much has changed since then, but the largest portion of the documents are web sites and bits of knowledge that I find online and then mirror so I have local access, and I don’t know what I have added to that collection.
I don’t think my backup habits are terrible: I keep my most important files in a small Perforce server with just one user, and the files are then checked out to the machine I’m working on, so a catastrophic data loss is unlikely here. I also I make a backup of my working data on “scalar” to a separate hard drive on the same machine twice a week. Once weekly I back up my working data from the MacBook to “scalar”. Every once in a while, I make backups of all the other things, and then copy them to USB hard drives or DVDs. I keep the DVDs and the USB hard drives in a water- and fire-proof safe and duplicates of the DVDs in my bank’s safe deposit box.
I really need to automate backing up the Perforce repository, though, and I also need to automate backing up the web server’s data. And now that I’m switching to Gigabit Ethernet, perhaps I should be thinking about about a network-attached RAID storage device.
I’m still in the process of restoring the web documents. Once that’s done I can examine the hardware damage. This is still a very shitty situation. I hate administrative tasks like this.
Update
When I opened up the case and tugged a bit on the cables, the hard drive came back to life. So it seems like I’ll at least have no data loss, but I don’t know how much I can trust that hard drive anymore. It’s an 80 GB Maxtor, by the way. I have a 160 GB Seagate and a 120 GB Maxtor hard drive left in “vector” which is quite enough.
If I were to buy a replacement hard drive. I could either get a few hundred gigs PATA for about $100, or I could get a 2×250 GB RAID 1 [2] Gigabit Ethernet device for about $300 or a 4×250 GB RAID 5 [3] Gigabit Ethernet and Wireless LAN 802.11b/g device for about $640… but I’d rather not spend that much.
