This morning, I shut down the notebook PC that I use for work and set off to meet a colleague. Upon returning, I tried to boot the system but nothing happened. Technically, something happened – but not what I expected – basically Windows would not boot and sometimes it reached the startup screen, sometimes it didn’t. Once or twice I’m sure I saw the once-familiar blue screen of death flash up for a fraction of a second before the PC reset itself. I tried a normal startup as well as last known good configuration, before finally I gave up and tried to recover the system using the Windows Server 2008 DVD but this couldn’t locate an installed copy of Windows to recover. What it would let me do though was get to a command prompt, where attempting to access drive C: returned:

The request could not be performed because of an I/O device error.

That didn’t sound good but I managed to run diskpart.exe and list disk told me that disk 0 was online. Moving on to try list partition told me that the two partitions I expected to see were there but it was list volume that really helped shine a light on the problem – the DVD drive and WinRE volumes showed as healthy but drive C: was reported as being a 110GB Healthy partition of type raw (i.e. not NTFS). At this point, I began to panic. Something had happened to the NTFS and that could mean lost data. I have a reasonably recent backup but the last couple of weeks at work have been mayhem and there was some stuff that I know I don’t have a second copy of.

I could call my company’s IT support number but it normally takes at least a day for a callback; I’d have to take the laptop to a “local” office (a 100 mile round trip); if a system won’t boot, the standard approach is to spend a very limited amount of time trying to fix it (probably none at all for people like me who run a non-standard operating system) before simply wiping the system and installing a new corporate build. That means going back to Windows XP and Office 2003 (which is painful when you are used to Windows Vista/Server 2008/7 and Office 2007), the loss of an activated copy of Windows Server 2008 Enterprise Edition (which is not exactly inexpensive) and also losing my data (the standard build has separate system and data partitions and my build does not… although now I’m starting to reconsider that choice).

I’m pretty sure that the root of this problem is a failing hard disk (after all, it is a “Western Dodgital“) but, without the tools to prove it, I’ve got a snowball’s chance in hell of getting a new one) and, to cut a long story short, when it comes to supporting my non-standard build, I’m on my own (at least unless I can prove that the hardware is faulty).

One of the podcasts that I listen to is “Security Now” and the hosts (Steve Gibson and Leo Laporte) spend far to much time plugging Steve’s SpinRite product. I’ve often wondered if it was any good but was not prepared to spend $89 for speculative purposes – this afternoon I decided that it was time to give it a try.

After paying up, downloading the software, extracting the ISO and creating a bootable CD, I ran SpinRite and performed what is referred to as a “level 2” scan. For the first 20 minutes, SpinRite ran through my disk finding nothing untoward but at the 50% mark it switched into “DynaStat” mode and started trying to identify lost data on one particular sector, slowly narrowing down the unrecoverable bits of the sector. Just this one sector took almost 5 hours and around 2000 samples but all of a sudden SpinRite took off again and finished up the rest of the drive in another 20 minutes. Even though the sector was marked as unrecoverable, a technical support conversation by e-mail confirms that this relates to the data, not the sector. With some trepidation, I restarted the computer, waited with baited breath and have never been so glad to see Windows start checking its disk(s). After a short while, chkdsk was complete and I was presented with a logon screen.

There’s nothing in the Windows event logs to indicate why my system failed to boot so many times this afternoon so it’s difficult to say what the problem was and whether it really was SpinRite that fixed it (although SpinRite did report the SMART data for the drive and there were a number of seek errors, backing up my theory that the hard disk is on its way out). What’s important though is that, as I write this post, Windows Server 2008 is 63% of its way through a backup and all seems to be well. I’m not quite ready to wholeheartedly endorse SpinRite – it does almost sound too good to be true – but, on the face of it, it seems to have recovered enough data on my disk to let Windows boot and for me to gain access to my system. That’s worth my $89 – although somehow I don’t see me getting that particular item through on my expenses…