On Wed, May 10, 2006 at 08:58:12AM -0400, Jon LaBadie wrote:
> On Wed, May 10, 2006 at 02:38:58PM +0200, Andreas_Hallmann wrote:
>
> > /dev/st0 is an Exabyte 8505 XL.
> 
> Are you really using st0, not nst0, for amanda.
> Or perhaps the messages are generic, not really
> referencing the drive name that was opened.
> 

The later, they are generic. I opened /dev/nrst0. Clearly.

> > The SCSI Specs speficy that "ASC/ASCQ: Write append position error" is a
> > more precise version of the message "media error".
> > 

> Only from googling:
> 
>       E6076 - Write append error
> 
>       Problem
>       If a WRITE or WRITE FILEMARKS command fails because the append
>       point was unreadable, this error occurs.
> 

Found that too, for an IBM tape drive IIRC.

This is my best candidate. My drive does read after write checks.

I would love to have a pointer to some scsi-generic explanation of thoose
ASC/ASCQ error codes. I only found code tables so far.

> Another google hits said this occurs when a write is attempted at EOT.

Yeah, also found that. Definitely not true here. Only wrote half a gig.

> And one said some "platforms" gave this error when a write was attempted
> in the middle of an existing tape file rather than at its end.

The term "platforms" must be kidding. These are SCSI error codes which are by 
definition
platform independent. Moreover reusing a tape we always do exactly that.
Or should it mean, we tried to write a tape file while another wasn't finished?
I don't believe amanda behaves that way.

aha
-- 
NetBSD: If you happen to have any problem with your uptime.