Category Image The tribulations of hard disk failure. 


Brad's PowerBook hard drive decided it had worked long enough. Here' an account of the tribulations Brad had in getting back up and running.  

My PowerBook is the core of my computing life these days. It goes virtually everywhere with me, and is the system I do all my work on. It contains my e-mail, my bookmarks, my documents, my local source repositories, my photos, movies I have made, and the music I listen to. It's my mailbox, TV, telephone, and stereo system all rolled into one device. So when it stops working, it's a big deal.

Late last week I suddenly had to reboot the system because it had apparently crashed -- an unusual situation in and of itself. I could hear that the hard disk was spinning and that the heads were scooting over the platters madly, but nothing was happening. As I had to give it the Mac equivalent of the three-finger salute (in this case, Ctrl-Command-Power), I decided to boot up off my iPod (on which I maintain a small, up-to-date OS X install just for such maintenance tasks) and verify that the hard drive was okay.

It wasn't. The SMART Status reported that the hard disk was failing. So while running the OS off my iPod, I decided to try to back up everything I possibly could across my home network to a Fedora Core 4 machine.

Which brought me to my first problem: the FC4 machine has a smaller hard drive than my PowerBook. I quickly decided not to back up everything, but to try to concentrate on two primary areas: my home directory, and my applications.

The home directory appeared at first to be easy. I'm running a system on my PowerBook which Apple calls FileVault, which automatically encrypts your home directory on the fly. This is implemented by making your home directory into a mountable encrypted disk image. The important thing here is that in essence, your home directory appears as a single file to the file system, meaning that backing up my home directory should be as easy as backing up a single file.

This, unfortunately, didn't work out as planned. Some of the data inside my home directory was unreadable, so trying to backup the entire encrypted file vault would fail about 8GB in (the image was about 12GB in size). Initial attempts to backup via Samba failed earlier (around the 4GB mark).

For those of you who have been fortunate not to have to deal with a failing drive before, one of the key things is to use it as little as possible while attempting to get your data off it. In my case, I had wasted a few hours in my first attempts to get the entire home directory disk image off the system. Not good. However, the drive continued to function for the most part, so I kept trying.

A direct file-by-file copy worked somewhat better, and in hindsight is what I should have tried the first time. Unfortunately, there were a few files that couldn't be copied in this way due to naming issues -- some files which had valid HFS+ filenames couldn't be copied through Samba to an ext3 partition. Fortunately, these were the exception rather than the rule, although in some cases I was worried about lost resource forks.

Things started moving across the network fairly smoothly, but the fates had decided to throw me another curve by sending a thunderstorm my way. My neighbourhood is currently serviced by a temporary power step-down transformer station built into a flatbed trailer, and it's been known to be a tad bit flaky now and then. It probably doesn't help that Ontario's power transmission utility, Hydro One, has had its professional engineers on strike for the past several weeks, but the end result was that for nearly an hour the power would brown out every few minutes -- enough to reset a computer which isn't on battery.

Fortunately for me, I had the foresight to plan for such things, and my FC4 machine is indeed on a UPS (the PowerBook, obviously, falls back to its built-in battery when the power fails). So both the sending system and the receiving system kept running. However, my network routing hardware would power down. I was nervous that the transfer would abort, however fortunately both neither system times out, and the transfer would begin again once the reset had completed.

I was concerned, however, that at any time a significantly longer power interruption could occur, so I moved my router off the grid and onto a portable power pack and DC->AC inverter I keep in my car. It was fun to watch as the power went down and up, only to have everything continue to work.

Eventually, the storm passed, and so did my luck. After backing up all the other directories in my home directory, I decided to back up my Library by zipping it up (as the Library directory contains some of the oddest filenames, and I was starting to get concerned about free space on my FC4 box). But part way through backing up my 2005 e-mail messages, the drive gave up the ghost, and stopped working completely.

(As a side note, only two days prior to this, I had archived all of my 2004 e-mail messages, which are safe).

Once the drive was dead completely, it was time to take a trip to the Apple Store. After over an hours wait to see someone at the Genius Bar (who verified my results), I found out that as my system was just over a month out of warranty, replacement with an identical 60GB, 4800RPM drive was going to cost me over $600 (Canadian), and was going to take 8 to 10 days.

Ouch.

Fortunately for me, Hal at the Yorkdale Apple Store Genius Bar is quite the decent guy, and pointed me to another Mac store in the area which would probably be able to replace my drive for significantly less money. So off I packed myself to CPUsed, where I was able to pick up a 80GB, 5400RPM drive for half the price what the Apple Store wanted. The one catch: I decided to install it myself.

Fortunately, this wasn't as bad as I anticipated, even though there are nearly 30 screws of different sizes and types that need to be removed in order to replace the drive. With the new drive in place, I was ready to restore everything.

Flashing forward to now, I've been able to restore everything, but have lost a few important items:

- Some movies I had made of my nieces,
- A few hundred e-mails from January 2005 until about two weeks ago,
- All of my applications and application settings.

I've been busy re-installing and reconfiguring everything, and haven't completely got everything back to where it was, but I'm at least 95% of the way there. Various backups I had on my .Mac iDisk turned out to be a big help, and to be honest having the bigger, faster drive with a from-blank OS X 10.4 install on the drive has made the system feel significantly more responsive.

The moral of the story? I need a better backup strategy. The most important items are already backed up to several different locations, but I need to come up with a system to backup some of the larger items, like iMovie files. A firewire hard drive may be in my future, along with more comprehensive network backups. 

Posted: Sunday - July 17, 2005 at 08:55 PM          


©