After just a couple of days I had the machine pretty much ready to replace my old T61 as my main customer demonstration machine. Except for one thing. Every now and again the machine would seemingly just lock up with the hard disk light on permanently. Sometime it would recover and sometimes it would require a hard reboot. I took to starting Resource Monitor in Disk mode as soon as the machine has booted to see if I could catch the culprit. You might think that my immediate reaction would be to blame Windows 7, but that was the last thing I was thinking. Since most of the Lotus stack that I had installed were builds and betas rather than gold releases my first guess was that there was a bit of dodgy code in there causing the problem. But some basic troubleshooting showed that they were in the clear. The problem sometimes occurred when nothing "unofficial" was loaded. I could even reproduce it in safe mode.
There seemed to be no rhyme or reason to the hangs. Sometimes it would go hours or even days without an issue but I couldn't trust it to use in front of customers. It was very frustrating having this Ferrari of a machine but not being able to know when it would freak out. Many people point to the indexing and pre-fetching in Win7 as being possible causes but I have had no problems on my home PC which is Win7 32bit so I was sceptical that this was the cause. It also gradually became apparent that even though the disk light was permanently on there wasn't that much disk activity going on, but the disk queue length was shooting up to 50 or 100. This would imply that two processes or threads were squabbling over disk access - but I could find no evidence of what they were. There was no fragmentation on the disk and CHKDSK was coming back clean.
Then out of the blue yesterday this error started appearing.
So it turns out that the problem was a failing disk drive. Not something I would expect on such a new machine. Hitachi doesn't publish a Mean Time Between Failure (MTBF) for the Travelstar 7K500 but it just goes to show that MTBF does not mean how long a drive will last but rather the average. So be warned and never be complacent. Drives can fail at any time. So make sure that you have a good backup. Fortunately I backup my important data to multiple locations (Home Network, Backup Thinkpad) and my really important data (customer presentations) to Lotus Connections Files/Lotus Quickr Places. So it will be very straight forward to rebuild this machine when my replacement drive arrives.
