[comp.sys.sun] Multiple dumps to /dev/nrst8 fail

kim@uunet.uu.net (Kim Kempf) (12/14/88)

I'm having a problem with cartridge tape dumps that used to work on SunOS
3.5 and now doesn't on SunOS 4.0.  The following:

	dump 6ucf /dev/nrst8 /dev/rxy0e
	dump 6ucf /dev/nrst8 /dev/rxy0f

all dumps fine with no reported problems.  restore will only work on the
first file saved.  Attempts to restore the other files on the tape (with
'restore is 2') will fail with the vague message:

	Tape read error: Error 0

Anyone else note such behaviour or am I just doing something wrong?

Kim Kempf, Microware Systems Corporation	...!sun!mcrware!kim

kim@uunet.uu.net (Kim Kempf) (12/22/88)

mcrware!kim@uunet.uu.net (Kim Kempf) writes:
>all dumps fine with no reported problems.  restore will only work on the
>first file saved.  Attempts to restore the other files on the tape (with
>'restore is 2') will fail with the vague message:
>
>	Tape read error: Error 0

I had many inquiries by others who have noticed this problem so it's worth
posting here.  This problem seems to have been fixed in the SunOS 4.0.1
patch tape "scsi" patch.  I think it was the "record length error should
be ignored" bug that caused the problem.

Kim Kempf, Microware Systems Corporation	...!sun!mcrware!kim

scs@uunet.uu.net (Steve C. Simmons) (12/23/88)

In article <855@mcrware.UUCP> mcrware!kim@uunet.uu.net (Kim Kempf) writes:
>I'm having a problem with cartridge tape dumps that used to work on SunOS
>3.5 and now doesn't on SunOS 4.0.  The following:
>
>	dump 6ucf /dev/nrst8 /dev/rxy0e
>	dump 6ucf /dev/nrst8 /dev/rxy0f
>
>all dumps fine with no reported problems.  restore will only work on the
>first file saved.  Attempts to restore the other files on the tape (with

Bug in 4.0.  Fixed in 4.0.1?  Who knows.  Workaround: to restore from
later save sets, do a 'mt -f /dev/nrst8 fsf <n>' to get close, then one
instance of 'mt -f /dev/nrst8 rsf 1'.  Now give the restore command.

The bug is in the tape positioning.  Apparently when you do the 'fsf', it
does not leave the head quite far enough down the tape to get past the end
of the tape mark.  The 'rsf' (record skip forward) gets it enough further
down the tape that 'restore' works.

Steve Simmons		...!umix!itivax!scs
Industrial Technology Institute, Ann Arbor, MI.

id@stl.stc.co.uk (Ian Domville) (12/30/88)

mcrware!kim@uunet.uu.net (Kim Kempf) writes:
>I'm having a problem with cartridge tape dumps that used to work on SunOS
>3.5 and now doesn't on SunOS 4.0.  The following:
>
>	dump 6ucf /dev/nrst8 /dev/rxy0e
>	dump 6ucf /dev/nrst8 /dev/rxy0f
>
>all dumps fine with no reported problems.  restore will only work on the
>first file saved.  Attempts to restore the other files on the tape (with
>'restore is 2') will fail with the vague message:
>
>	Tape read error: Error 0
>
>Anyone else note such behaviour or am I just doing something wrong?

A few weeks ago, I sent out a similar request on Sun-spots. I received
back several very helpful replies, and I now have a solution to the
problem.

I actually had two problems.  The  first was to do  with tape length.
SunOS 4.0 was supposed to handle tape length correctly. In practice, I had
to specify 575' for a 600' tape. Other people on this site had the same
problem. One reply told me it appeared to vary from make to make. (I was
using 3M cartridges.)

The second was the same as Kim's.  The cause  is that the  new dump
program puts an extra EOF at the end of each file. Consequently, you must
restore files 1, 3, 5, ... to avoid the extra EOFs.

One reply suggested an alternative of dumping the first, rewinding,  mt
fsf 1, dump the second, etc. This gets rid of the extra EOFs,  but sounded
too much like hard work to me.

- Ian Domville

PHONE  	: PHONE	: +44 279 29531 x2576
POST	: STC Technology Ltd., Old London Road, HARLOW, Essex, CM17 9NA, UK.
ARPA	: id%stl.stc.co.uk@cs.ucl.ac.uk
JANET	: id@stl.stc.co.uk
UUCP	: id@stl.UUCP


-Ian Domville (id@stl  ...!mcvax!ukc!stl!id  +44-279-29531 x 2576)

pvo1478@oce.orst.edu (Paul O'Neill) (01/07/89)

There are 2 tape marks after every dump.  Skip files accordingly with mt
fsf x and there is no problem.  See sun-spots v6n209.

Paul O'Neill                 pvo@oce.orst.edu
Coastal Imaging Lab
OSU--Oceanography
Corvallis, OR  97331         503-754-3251