Wednesday, February 16, 2005

I hate Western Digital

"...based on the lightning doesn't strike twice principle I should be right for a while now."

That was me almost exactly two months ago. I'd just had a hard drive failure in my server and lost my source code version control repository. It was an annoyance but I could get by since I still had the latest version of the Sydney source code on my working machine. I took it as a warning to improve my backup procedures but reasoned that statistically speaking I should be safe from hard drive failures for a while.

Statistically speaking, I'm an idiot.

The second hard drive that was in that server (another Western Digital 120GB IDE drive) has just failed. While I had some warning that the first drive was going to fail this one died suddenly before I could pull anything off it. On it was my version control repository (again), my build server virtual machine, a bunch of testing virtual machines and most importantly several years worth of real world test data that I have been gathering to feed into Sydney.

And what of my improved backup procedures that the first drive failure prompted me to put into place? Unfortunately they're still on the drawing board. I have a old machine ready to use as a backup server that I'll keep at my house but I haven't finished putting it together yet. Another week and it would have been ready, I've just been too busy with the moves and technical support jobs of late to devote more time to it.

The good news

The good news is that by going into Windows safe mode and using xcopy repeatedly it appears I was able to recover some of the files. I've retrieved my version control repository, an older version of my build server virtual machine and maybe 15% of the test data archive. The drive has now completely died, emitting all sorts of scraping, clicking and clanking noises if I try to use it.

So once again, through good fortune rather than good planning, I seem to have averted disaster. I'm not going to tempt fate again though. As I said last time

"Of course if lightning does happen to strike twice I'll have to be very wary. These things happen in threes you know."

The backup server will be up running at my house before the end of the working week.


At 12:35 PM, Anonymous Anonymous said...

That *REALLY* sucks.

i guess you were able to recover a fair bit of it...some kind of silver lining, eh?


At 1:54 PM, Anonymous Anonymous said...

Something similar happened to me and I was able to get a lot of data back with RecoverLost Data from StompSoft. Costs USD 39.99

At 9:37 AM, Anonymous Sebastian said...

Did you think about setting up a low cost IDE RAID-1? That way you might not even have to resort to your backup during a drive failure. I think simple RAID controllers are not that expensive these days. Unfortunately all I could dig up was a pretty dated review about RAID controllers from November 2002.
PS: Good luck with Sydney

At 6:15 AM, Anonymous Anonymous said...

I have a RAID mirror at home and it was easy enough to set up (plug in controller, it copies from one drive to another, forget about it). If I yank a drive as a test it gives me warnings on startup that I'm driving without a spare but otherwise just keeps running, and if I put the drive back it resyncs things. Easy and pretty cheap - the controller was quite inexpensive, most of the cost is the second drive which is no big deal.

At 8:03 AM, Blogger Lachlan said...

I would look at a low cost IDE RAID system but part of the purpose of the backup server is as an offsite backup. The backup server will be kept at my house while the main server will be at the office.


Post a Comment

<< Home