[Info-vax] analyze/disk errors

[Stephen Hoffman]

On 2013-12-07 00:41:46 +0000, David Froble said:

> As mentioned earlier, while back in the day the file system handled bad 
> blocks, thus BADBLK.*, today's disks handle them transparently.  That 
> part of the VMS filesystem just isn't needed anymore.

SCSI disks don't handle the errors transparently.  Sure, once the 
revectoring actually happens and the blocks are shuffled, the disk is 
once again logically perfect, but between the time when the error is 
detected and when the block is rewritten and particularly in the 
absence of HBVS, the contents of the bad block is returned to the user 
with a PARITY error, on the off chance the user can get data out of it.
> 
> I also would never have tried to mess with BADBLK as a file, because it 
> really isn't a file, it's part of the structure placed on a disk that 
> the ODS-* file systems require.

It really is a file.  It just gets some funky block mapping that 
ANALYZE /DISK should really either just flag as an indication of a disk 
bad block problem that's currently active, or should ignore.

> Yeah, get Steve those bad disks, don't know how else to test for bad disks.

Controllers add a whole 'nother layer of weird here, too.


-- 
Pure Personal Opinion | HoffmanLabs LLC