Random Chaos

Smart HDD Failure Prediction

My older computer which is now a headless box (but still handles my browsing/email/IM and also acts as a gateway machine for my home network) didn't come back up on the network after a BESCOM induced shutdown, on 6th afternoon.

So hooked up my monitor to it and was greeted with this screen.

(Bit of trivia S.M.A.R.T. - Self-Monitoring, Analysis and Reporting Technology, and that was probably the first time I saw a BIOS message asking me to hit F1 for anything other than "keyboard error"!)

I bought the disk in UAE back in 1999, and it has housed my primary operating system installation since then, and has lived in 2 different computers I used before finally settling down in this machine in early 2001 once I got a phone line at home and could work from here. So wasn't too surprised since it had lasted a good 6 years.

Called up a couple of friends to find out about the accuracy of these kind of warnings, and was told to use the smartmontools to get a more detailed report, which read thus...

 === START OF READ SMART DATA SECTION ===
 SMART overall-health self-assessment test result: FAILED!
 Drive failure expected in less than 24 hours. SAVE ALL DATA.
 Failed Attributes:
 ID# ATTRIBUTE_NAME          FLAG     VALUE WORST THRESH TYPE      UPDATED  WHEN_FAILED RAW_VALUE
   5 Reallocated_Sector_Ct   0x0033   017   017   020    Pre-fail  Always   FAILING_NOW 416

So went out that evening and bought myself a 160GB Seagate harddisk which cost less than what I paid for my 16GB Quantum 6 years ago!(and on a whim decided to buy an extra 512MB of RAM for my iBook, which cost much less than the official Apple rates, and my iBook doesn't crawl while opening iPhoto. Ironically I have more RAM on my laptop now than on my desktop.)

In an ideal world, getting the new hard disk and moving the data over to it would have been the end of the problem. But instead it was just the beginning of the real problem.

Once I'd rsynced all the data on the old disk to the new one and taken out the old disk and tried to boot up the system.. it wouldn't see the new disk completely. My motherboard came with a BIOS dating back to 2000, that wasn't capable of handling a 160GB disk. And the floppy drive on that machine seems to have died a silent death from disuse and loneliness. And my DSL connection decided to play dead that night and most of the following day. Googled and found a couple of rescue CDs with bootloaders which bypassed the BIOS but most of them were too big to download on my slow GPRS link.

Discovered Gujin, which is a pretty impressive concept bootloader. I created ISOs that worked fine under Qemu and bochs, but just wouldn't work on the machine I wanted it to. Ended up with a nice collection of coasters and still no working system. The author was kind enough to correspond pretty regularly and try and troubleshoot my problem. Merci Etienne!

Finally found that the BIOS could be updated from a CDROM drive with an image of a DOS boot diskette containing the BIOS update. So download a DOS boot disk, got rid of all the unwanted extras and put two BIOS updates onto the disk and created a boot-floppy-cd image and since their website mentioned that the stable BIOS update only supported a disk of 120GB, I decided to go with the beta version which luckily detected the entire 160GB disk.

But the system still wouldn't boot for some reason lilo/grub wouldn't work when I installed them on the hard disk itself. Finally got tired of it and installed grub on another disk on the same machine and set it to boot the OS on the new disk.

Finally more than 2 days after seeing the warning, my system was up and running again. I sure hope the SMART System on my Seagate harddisks are as good as the Quantum one. Maybe it's time to move onto redundant/fail-safe storage.