On Sat, 1 Mar 1997, Benoit MARTEL wrote:

> Unfortunatly, it seems it may be worse. Now, when I run fsck again, I get 
> the error "/dev/rsd0a: CANNOT READ: BLK <nnnnnn>" <nnnnnn> being a 
> number. I ran fsck many times and about half the time, it was successful 
> and the other half, it gave the above error msg with a different number 
> each time.
> 
> I feel somewhat unsafe using the system as it is now (but I will because 
> I need to do some work). What should I do to "recover" properly from this 
> strange situation? At worst I can backup my personal files and reinstall 
> everything. And what did I do wrong so I can be careful for it not to 
> happen again?

First, toss the netbsd kernel and switch to any "generic" kernel that uses
the sbc SCSI driver.  The above are signs that you're running headfirst
into an ncrscsi compatibility problem....  Using the sbc kernel should
stabilize things.  Then, reinstall all of the distribution packages except
etc12 (don't want to wipe out your configs, after all).

> Because user errors often produce unpredictable results, the user
> should try to avoid them.
> 
> IBM MVS/XA System Programming Library.

I _really_ hope you're kidding about this one....  Sounds a lot more like
(insert your favorite Microsoft product here) to me....  :-)

Later,

 /---------------------------------------------------------------------\
|David A. Gatwood             And Richard Cory, one calm summer night,  |
|davagatw@mars.utm.edu      Went home and put a bullet through his head.|
|dgatwood@globegate.utm.edu          --Edwin Arlington Robinson         |
|-----------------------------------------------------------------------|
|http://globegate.utm.edu                  http://www.utm.edu/~davagatw |
|http://mars.utm.edu/~davagatw             http://www.nyx.net/~dgatwood |
 \---------------------------------------------------------------------/