Hi,

I have four recently aquired refurbished Western Digital Caviar 250 GB
7200 RPM EIDE drives in a SUSE 9.3 linux box. When I use the linux
command "badblocks -s -w -o hdX_badblocks.txt /dev/hdX1", I get this
result (by typing "wc -l hd*.txt"):

  4480 hda_badblocks.txt
  4096 hdb_badblocks.txt
  3328 hdc_badblocks.txt
   435 hdd_badblocks.txt

(Each number is the number of reported bad 1024-byte blocks found
during the test on each drive.)

These tests are still in progress, and when they complete, I'm going
to rotate the drives to adjacent spots on the IDE bus to rule out
wiring issues.

Question #1: assuming I get the same (or nearly the same) result when
I move the drives around, how many bad blocks is too many? Do I have
bad drives?

Question #2: when I assemble these drives into a raid5 volume, can I
be assured that these bad sectors will be avoided? How can I verify
this?

Thanks in advance!

---

Clarification of Question by spblat-ga on 23 Jan 2006 16:45 PST

Note to self. 

umount /dev/md0 isn't enough.  Don't run badblocks while the raid is
still running.  It will try to resync a volume undergoing constant
change, and it could take centuries to finish, not to mention the
possibility that badblocks will think it sees errors when in fact the
raid system is madly trying to fix the mess being made by badblocks.

raidstop /dev/md0

Please don't bother answering for now.  I'll run the tests again and
post the results.