[ILUG] Hard Drive Resurrection?

[Thomas Ribbrock]

Hi all!

Just had a strange experience with two SCSI hard drives I recently got to
experiment with... (well, maybe it's not so strange to someone out there who
can explain it to me... ;-) ).

Both drives were declared as dodgy by their previous owners and sure enough,
both showed plenty of errors like this one:

Info fld=0x261727, Current sd08:21: sense key Medium Error
Additional sense indicates Unrecovered read error
scsidisk I/O error: dev 08:21, sector 2496262

So, just for the heck of it, I ran "mke2fs -c" several times over them (up
to twenty times on one of them) and lo and behold: All errors gone.

So, next step, I found this most helpful resource in Linux Journal:

http://noframes.linuxjournal.com/lj-issues/issue33/0193.html

"Disk Maintenance under Linux (Disk Recovery)"

Following that, I ran dumpe2fs on the drives, and used the output to run 
"badblocks -wvo FILE /dev/xxxx block_count", i.e. a full badblocks run with
write and read tests (Warning: Kids, don't try this at home - at least not
with a drive that contains data you'd like to keep!). Result: Nada. Zilch.
No errors. One of the drives gave a single error on a subsequent mke2fs, the
other was entirely quiet.

Now - does anybody happen to know how these drives could be "self healing"?
I mean, I probably still won't use them for anything really important (at
least not until after at least a dozen more badblocks and other stress
runs...), but I'd be curious to know how this works...

Cheerio,

Thomas
-- 
-----------------------------------------------------------------------------
      Thomas Ribbrock    http://www.bigfoot.com/~kaytan    ICQ#: 15839919
   "You have to live on the edge of reality - to make your dreams come true!"