780 Crashes

[Glenda Stubblefield]

I am having a problem with a VAX 780 running Mt. Xinu 4.3.  
We have RA81 drives and UDA50 disk controllers.

On the average of once a month (since May of this year) I have
lost my /usr filesystem because of a system crash.  Since 7 Oct
this has happened three times, the latest was yesterday evening.
I have had the CE here repeatedly and he has cleaned, switched, and
replaced cables.  The latest error message, from last night, is
as follows:
   machine check 3: cp cache par fault
           cpues 10083 upc 24d va/viba 80557598 dreg 20000
                 tber 7e81 0
           timo 9c58 parity ff7f sbier 8002 pc 80009c5a psl
                40c0009 sbifs 2000000
   panic: mchk
   syncing disks... machine check f: ib cache par fault
           cpues 10083 upc 27d va/viba 8001f598 dreg 82
                 tber 7c81 0
           timo 1f594 parity bf7f sbier 8002 pc 8001f595
                psl 40c0000 sbifs 2000000
   panic: mchk

   dumping to dev 901, offset 58688
   dump to ra0 succeeded

After these crashes I cannot get thru a fsck because the /usr
file system comes back with BAD SUPER BLOCK:  MAGIC NUMBER WRONG.
If I continue the fsck manually everything will go into lost+found,
and it has been easier for me to just restore off my full system
and incremental dumps rather than have that ENTIRE file system
in lost+found.

I'm not getting anywhere with the CE, and I wonder if there's
something that someone else out there can see that I'm not
recognizing.  I'm identifying this as a hardware problem, maybe a
memory problem?  Am I wrong?

I'd appreciate any thoughts anyone might have that would help me
out of this mess.

You can write directly to me rather than to the mailing list, if
you wish.

Thanks,

Glenda Stubblefield  stubblef at almsa-1
CSDA
St. Louis, MO