Hi!
Trying to kill the keyboard, [EMAIL PROTECTED] produced:
> Hello Onno,
> Sunday, November 14, 1999, 7:20:13 AM, you wrote:
> OG> 2. In case the tape works there are fatal errors on many tapes: Sometimes
> OG> the drive attemps to write on the same position over and over as if the
> OG> tape were damaged and finally stops with a fatal io error. On some tapes
> OG> this happens immediately after starting the backup, on others after the
> OG> drive has worked for a while.
[but they don't show up under DOS]
I pasted in some of the relevant lines (or what I think they
are) from below ...
> OG> (ftape_write_segment) - write error, retry 6 (1801).
> OG> Nov 14 11:13:10 diden kernel: [071] 0 ftape-write.c (write_segment) -
> OG> warning: 1 hard error(s) in written segment.
> OG> Nov 14 11:13:10 diden kernel: [072] 0 ftape-write.c (write_segment) -
> OG> hard error in segment 1801.
We see here that there *is* a bad spot on the tape.
> OG> (zftape_write_segment) - Hard error writing segment, trying to recover.
> OG> (ftape_write_segment) - write error, retry 1 (1801).
> OG> Nov 14 11:14:04 diden kernel: [079] 0 ftape-rw.c (ftape_start_tape) -
> OG> failed to reposition.
And another one, and it seems that the recovery routine is
somewhat confused that more bad spots appear while it's
trying to sort the last one out.
Worse: It tried to reposition the tape but that failed, too!
Now, there's little at that point that ftape knows to do, so
it dies horribly.
It may be that the recovery routine under DOS is better
tested or working in a different way (maybe just ignoring the
errors!) and does not show them thus.
> OG> 3. I downloaded ftape-tools (latest stable version). However ftformat does not
> OG> work. It tells something about unknown API.
You might have luck with ftape 3.04d (which is in the kernel,
IIRC), because the only way to help ftape (short of hacking
the code where it's suboptimal) is to tell it beforehand where
they are. That is done by formatting (or at least checking
for bad spots), but I'd go for formatting if I could.
> OG> Upgrading to the latest unstable tools wasn't possible because they didn't
> OG> compile. The error messages were:
> OG> /lib/libc.so.6: warning: multiple common of `_nl_domain_bindings'
> OG> ../../intl/libintl.a(dcgettext.o): warning: previous common is here
That seems to be a problem local to your computer, but then I
may be completely wrong!
> Here's the steps to fix the ftape problem:
Agreed, IF you have the money. And are not bound to that
hardware for other reasons.
> 1. Buy one Adaptech SCSI host adapter - about $60 US
Any odd SCSI host adapter that meets the specs of the tape
would do, or not?
> 3. Get rid of floppy tape drive and concept.
Ack. They are a terrible hack, but then these are everywhere
(say BIOS and 1024 heads, say 502 MB, say Windows(most any
version), say Intel Architecture, say Winmodem, say charmaps,
and ... and ... and ...)
-Wolfgang
