Oliver -
I must have missed your first post.
The problem you are seeing **should** never occur. The fact that you
are getting beyond the NULL data zone on the tape to actually see old
data is controlled by the DDS drive's firmware, not something in the
Linux kernel drivers. This tells me that your tape drive may be on the
way out.
It could be that when you were writing this tape, the close was
interrupted by a power off or SCSI bus glitch that caused a drive reset
before the drive could write the NULL data zone (EOD marker).
Just so you understand, when you are writing to a DDS drive (and most
other intelligent SCSI tape drives), you should be permitted to write
data at two locations - Beginning of Tape (BOT) and End of Data (EOD).
The way that a drive identifies BOT is obvious. Determining EOD is
slightly different, as it requires a marker to let the drive electronics
know that you've reached the EOD position. The way this is done on DDS
drives is through the writing of an arbitrary sized NULL data zone that
is easily read by the unit during a fast seek. If this marker is
somehow absent or missed, and additional data exists on the media, the
drive will continue seeking until it DOES locate the EOD marker, or
until it reaches (End of Tape) EOT. The fact that you appear to be
reaching another EOD marker implies to me that the creation of the
marker (the NULL data zone) failed for some reason on the latest write
that you performed.
Do you have other tapes to test this problem against? I would recommend
a series of file sets written using the non-rewinding device, followed
by an overwrite test from BOT.
--
Tim Jones [EMAIL PROTECTED]
Vice President http://www.estinc.com/
Enhanced Software Technologies, Inc. (602) 470-1115
"The BRU Guys"
[EMAIL PROTECTED] wrote:
>
> Hi,
>
> I have already posted this, and got no response. Is it really
> so big question?
>
> I have a HP-SCSI-Streamer. Until now, I have written more than one
> file on my types. this worked very well. I used the command
> "mt -f /dev/nst0 eod" to position the tape to the end an wrote another
> file on it. For searching I used "fsf 1" and so on. Now I overwrote
> the whole tape. I rewinded the tape an put I file on it from the
> beginning of the tape. Until some days, when I used "mt eod" the
> tape was positioned after this file, but now it is positioned after
> some more files (the old ones which where on the tape befor). IMHO the
> old files should normaly not be acessible, which also was the behaviour
> some days ago. So what is happening now? The only thing which I have
> changed and which might have to do with streamers, is the kernel.
> Does anyone have a idea?
> What do I have to do to get the tape positioned after the last
> valid file?
> I use Kernel 2.2.12 and mt-st-0.5b.
>
> Thank you
>
> Oliver
> --
> ,,,
> (o o)
> --oOO-(_)-OOo-- [EMAIL PROTECTED] - [EMAIL PROTECTED] -
> [EMAIL PROTECTED]
