Adaptec 2400A RAID controller corrupting data (4.8)

[Matt Staroscik]

I have started building myself a new file server, and at the heart of it 
are 2 Maxtor 160GB drives in a RAID-1, using the Adaptec 2400a. 
Unfortunately I am having some kind of issue with data on the array getting 
corrupted.

During disk activity (like makeworld, cvsup, rm -rf /usr/obj/*) I get 
kernel panics like this:

dev=#da/0x20006, block=54608, fs=/usr
panic: ffs_blkfree: freeing free block

Upon rebooting I have to fsck -y to repair the damage.

I have also seen truncated file or file missing errors during builds, as if 
the files I cvsupped weren't written properly to disk.

I am conducting these tests on a plain vanilla 4.8 install from CD. Other 
system details: ECS L7VTA motherboard (KT400 chipset), AXP 2400+, 512MB DDR 
RAM, vr NIC on mobo, dc NIC added in, Matrox Millennium II video card (had 
the same problems when I had a Cirrus card in too).

All these data corruption problems and panics go away if I put a drive on 
the motherboard's own IDE controller and work on it.

Weird thing is, this box was working GREAT just a day or so ago. I had 
multiple successful builds on the RAID array. But something has gone south.

I am going to check cables etc. of course, but if anyone has insight... 
please send it my way. :) There are a lot of things this COULD be and they 
will take a LONG time to test. For example, those 160GB drives take about 
10 hours to low-level format, so I am hoping to solve this without 
rebuilding the array from SCRATCH. (and as it reports "optimal" status I 
don't feel like I should have to, but...)

Many thanks in advance. When/if I solve this I will post a followup...

Cheers,
Matt