I got conned into buying a micropolis
scsi hd.
The first three
arrived dead and had to be sent back (with a 3 month delay each time). The third I had to
low
level format before it would work at all. It lasted
a total of 3 months before it too was sent back.
Oddly enough, when I did a postmortem on it before
sending it back, only the areas with lots of writes
had problems. High activity areas with only reads
had no errors at all. (I guess trying to swap on it wasn't a good idea.)
When I got the replacement drive (6 months later), I
just used it to cache cdrom's. The drive is still
functional, and has not run out of replacement sectors yet. Did I finally get a good drive, or is it just
holding up under the read I/O strain? Idunno. I'm not
gonna test it either. I probably got that drive (the replacement) around 8/97; it is still going strong in 2002!
updates:
March 2002: I recently deleted some stale cdroms from that drive
and copied some new ones on. Untouched volumes are now showing bad sectors. The directory inodes seem
to be going bad first. (Bad sectors in inodes, oh my!) I suspect the drive would be fine if I had turned
off atime--the nightly updatedb must be killing it. Too bad the OS it is currently attached to supports neither bad sector support nor the noatime mount option.
May 2002: The drive has died, of course. It makes strange clicking
noises periodically, and does not respond to I/O requests.
2017: I still have some of the disassembled pieces of this drive on my desk. It's a monument to the hugeness of disks back in the day!