[CLUE-Tech] Problems with 3ware 7006-2 RAID controller?

[Jim Ockers]

Hi all,

I was wondering if anyone else on this list is using a 
3ware Escalade 7006-2 PCI RAID controller?  I bought one a
couple of weeks ago thinking that it was going to solve my
storage problems for good.  Also the price was right, and
it's _real_ hardware RAID, not fake hardware RAID like some
other vendors *cough* *Promise* *cough*.

Anyway I have two 120GB Maxtor drives, set in a RAID1 mirror.
I have the tw_cli 3ware utility that lets me query the array
status and drive status.  Everything's OK, according to tw_cli.

However, I seem to be losing files.  My system logged the
following, and similar ext3_readdir errors:

Nov 14 21:45:06 tahoua kernel: EXT3-fs error (device sd(8,3)): ext3_readdir: 
	bad entry in directory #344065: directory entry across blocks - 
	offset=572, inode=17760284, rec_len=26988, name_len=98
Nov 14 21:45:41 tahoua last message repeated 3 times
Nov 14 21:47:02 tahoua last message repeated 2 times

lots of times.  A reboot and fsck caused the system to drop to
manual fsck mode because there were too many errors.  I am the
sort of user that has to answer 'y' to every question asked by
e2fsck, but this time there were some 250 "questions" .  Needless
to say a bunch of files got put in /lost+found and the filenames
are lost, so I am stuck trying to rebuild the system manually.

Anyone else have filesystem corruption apparently caused by a 3ware
RAID controller?  I'm on Red Hat 7.3, newest up2date kernel 2.4.20-20.7,
newest 3ware driver, firmware, BIOS, etc. etc. etc.  System is a P3
500MHz CPU, 256MB RAM, Intel 440BX chipset, nothing wrong with it
AFAIK.

The drives and 3ware card are brand new, and in my opinion there's
no excuse for this kind of behavior from the 3ware card.

Any ideas?

-- 
Jim Ockers, P.Eng. (ockers at ockers.net)
Contact info: please see http://www.ockers.net/