?

Log in

No account? Create an account
Dead hard drive - Alex Belits
abelits
abelits
Dead hard drive
On October 26 last year I have seen some hard drive errors on my home desktop, and decided to replace its 250G Maxtor drive, then merely 8 months old. Unfortunately, deciding and actually doing things are not the same, and I have forgot about that incident, until March 31, when the same problem happened again. This time it looked like drive ran out of replacement sectors, so errors were more numerous, and included things like /sbin/rc with obvious consequences.


Booting from Knoppix confirmed that drive is still mostly readable, but any hope for recovery to a bootable condition is lost, so I have moved this drive to my second box, and started copying my home directory while buying a replacement drive. Unfortunately, second box had a small drive, and tar.gz file quickly filled all available space -- I didn't even intend it to be anything to rely on, just a last-resort copy for the case if the drive will die while I will be making the copy on then-nonexistent new drive.

While this copy was being made (that was morning April 1st), I went to CompUSA, and bought a 300G WD drive, brought it home, partitioned/formatted it in a manner similar to the original drive, and started copying /home and /etc. At that point I have noticed that despite all drives being outside the case, in a cool room, failing Maxtor is unusually hot. I have placed a fan on top of the drives to provide some airflow, (not that it helped much), and most of the files copied without a problem -- errors happened to be in things like downloaded sources that I compiled things from, so I thought, it's nothing that cvs checkout won't fix. /etc that contained my HTTP server's configuration and DNS, wasn't a problem either. The rest looked broken, and I didn't even dare to try to export/copy MythTV stuff. I have left copies on the future /home filesystem, so now the problem was "just" to rebuild the system. "Rebuild" because the system was Gentoo, and it was the only Gentoo box that I had. And it was a system that accumulated a lot of stuff since its original installation a year ago. I have backed up portage configuration, downloaded and burned a new installation CD, then moved the hard drive to the old box, to install on the target system, what is kinda a point of having Gentoo.

Gentoo installer did not make itself any more lamer-friendly since the last time I have seen it, so it didn't ask me, if I want to reformat my /home with backups, or do some other inane thing with it -- base installation from a binary tarball, and building a kernel was as smooth as with any other system. At that point, finally a transition to udev was made. Then, of course, it was time to compile the enormous pile of packages listed in my world file. And that was long. Really long. And in the middle of it, the current xine package tried to find Xv in /usr/X11R6/lib after it was already moved to /usr/lib, crippling video playback in unimaginable ways, and I had to rebuild MythtV database from scratch, however in the end the system was restored to the same content of the world file, plus MythTV reinstalled from CVS, plus my HTTP server reinstalled, DNS zones restored, CUPS reconfigured, and iptables configuration copied from backup. Gnome picked up configuration from my home directory, and I have finally added all the Gnome automounter stuff (have you noticed that I have configured kernel for udev? Without that it wouldn't work, and documentation wasn't exactly clear about that). Dual-screen stuff (under binary Nvidia drivers) worked fine, so everything was supposed to be fine. OpenOffice was still compiling, taking insane time to complete, and I expected it to work just like the rest of the packages.

Then I have started OpenOffice. And it probably was perfectly usable -- for someone who knows Chinese. When I was installing languages packages I have configured LINGUAS variable starting with Chinese, and OpenOffice ebuild decided that this is the primary language, so Chinese -- and only Chinese -- menus were built. Hours later, OpenOffice was rebuilt for English, and that was the end of it.

Except that at this point it was already April 3rd, and I have wanted to take the last look at the failing drive. I have reconnected it to the desktop as a slave, turned it on, and after a second overcurrent protection kicked in. Disconnected the drive, and everything is back to normal. While carrying the drive, I have slightly turned it in my hand, and the drive produced the sound that left no possibility for a doubt -- heads were scratching the platters.

I have rescued my home directory in the last two days of the drive's life. And at that point the drive was 13 months old, running in a well-cooled case, along with other, older drive that never had a problem. Excuse me if I will not ever use Maxtors for anything important.

Tags:

2 comments or Leave a comment
Comments
dk379 From: dk379 Date: April 19th, 2005 07:23 am (UTC) (Link)
ever considered using RAID1 on your home boxes?
abelits From: abelits Date: April 21st, 2005 11:22 pm (UTC) (Link)
Considered, but decided that I would rather keep backups on my second box. And then forgot to schedule them.
2 comments or Leave a comment