Memory Loss

My main hard disk that stores /home and most of my important stuff recently failed. I went to save a file to a directory, only to find the directory missing. After seeing “IO error” all over the logs, I thought rebooting was needed to clear up a buggy controller or driver. Big mistake! The first sector, the MBR where the boot loader and partition table live, was already bad, making the drive unusable by the OS.

Lesson One: when you notice IO errors, immediately begin copying files to a safe disk.

I reboot, and at this point, I see my OS telling me “drive is unusable.” So I freaked out, thinking about the 16 months of digital pictures, years of e-mail, thousands of hours of source code all lost. I decided I would recover the drive no matter what the cost. I called half a dozen places like OnTrack and Drive Savers and got the bad news: with an 80 gig drive, it’s gonna cost in the range of $1000 to well over $3000. I felt sick, very sick.

Lesson Two: do your research and don’t let people take advantage of you.

So I keep hearing about and seeing web sites for data recovery software that doesn’t require the expensive trip to the clean room. In particular, I find software from R-TT that supports ext2 and is good quality. Unfortunately, the open source programs out there like e2extract are a little scary at alpha level and no documentation.

I was able to recover my data! And I didn’t even have to put my hard disk in the freezer (I’m speaking literally — this apparently works on drives with heads crashed on the platter). It turned out the disk was covered in bad sectors, but I still got 99% of my data back. And I learned a few things about super nodes and block groups along the way.

Lesson Three: backup your data! It’s not “if” your drive will fail, it’s “when.” Hard drives last on average 3-5 years. With 7200 RPM drives, the heat build-up makes them more likely to fail.

So now I am shopping for a RAID-5 controller card with 4 channels. The ATA RAID cards from Promise look very attractive. Maybe a bigger case with lots of fans too.

One Response to “Memory Loss”

  1. Rob Stevenson Says:

    Is RAID the best way to go for circumventing HD problems? I’m wondering about the differences between having a good disk checker (to look for bad clusters, etc) and a good backup program and going with a full RAID setup. Personally I’ve never had a HD crash on me YET. I also don’t have a good disk checker nor a backup program of any kind. I think I’m just asking for my HD to crash tonight.

    Now, with a RAID, if set up appropriately, you can just swap out a bad drive for a good one, lickety split. With a good backup program, reinstalling things isn’t that big of a deal either. Hmmm.

    Of course, with a backup program you’ll need some kind of media to back all your data on. CD-R’s are fast becoming the “Floppy disk of the decade” since they only hold 650-700MB a piece. DVDs look to be a better solution at what, 4.7GB? Can you even buy a tape drive anymore? Would you want to?

    Maybe a RAID array is just what the doctor ordered.

    Any thoughts?

Leave a Reply

You must be logged in to post a comment.