Article 168370 of comp.os.vms:
On 20 Mar 1997, W.F.J.Mueller wrote:

> One major problem with those disks is that the VMS bad block replacement
> simply doesn't work. As far as I remember the drives had to be configured
> with `write cache off' and `bad block replacement off'. The former to
> ensure proper write ordering of file system meta data, the reason for
> the later I don't recall. Now, once we had a file with a hard read error

VMS needs to turn off the drives block replacement so it can write
an `error' to the replacment block.

If the drive was left to it, it would trash the bad block, allocate
a replacment, then copy the garbageed data to the replacment with 
correct CRC... So you read the trash, and have no indication of a
problem. Or to put it in MSCPese, SCSI does not have a force error
 header bit. VMS uses a WriteLong to write the recovered data with a zero
CRC, so you will gwt a hard error on read. A ReadLong gets the CRC back,
and if = 0, a `force error' is returned.

The turning off of the write cache is more likely due to the drive
screwing up on a `ForceMediaAccess' is the data is in the cache.

~Paul