Package: dump
Version: 0.4b41-2
When I dump a filesystem which results in an archive file larger
than 0x7fffffff, the restore fails after transfering about about
2GB of data.
The command used to create the archives was
dump -0 -f /mnt/sda2/name /mnt/hdaX
which in each case completed without error.
I produced the following archives:
debian:/mnt/sda3# ls -l /mnt/sda2
total 14099291
-rw-r--r-- 1 root root 2777466880 2006-04-26 03:26 DebianEtch.dump
-rw-r--r-- 1 root root 1579960320 2006-04-26 03:16 ubuntu5.04.dump
-rw-r--r-- 1 root root 2081443840 2006-04-26 03:20 ubuntu5.10.dump
-rw-r--r-- 1 root root 2428569600 2006-04-26 03:32 ubuntuDappaBeta.dump
The two smaller "ubuntu" dumps can be restored without error. However
when I attempt to restore either of the two larger archives I get an
error after some time has elapsed as follows:
debian:/mnt/sda3# restore rf /mnt/sda2/DebianEtch.dump
Tape read error while restoring
./usr/share/texmf-tetex/fonts/tfm/public/pxfonts/pcxi.tfm
continue? [yn] y
Tape read error while trying to resynchronize
continue? [yn] n
debian:/mnt/sda3#
The symptoms are reproducible.
I suspect there is a problem with using a signed 32 bit integer
in calculating offsets into the archive somewhere.
I am running Debian Etch, netinstall, daily build from April 17th:
debian:/mnt/sda3# uname -a
Linux debian 2.6.15-1-686 #2 Mon Mar 6 15:27:08 UTC 2006 i686 GNU/Linux
I have reproduced the fault using dump/restore on Ubuntu Dapper Drake,
so I don't think the fault is specific to Debian.
The hardware is a Fujitsu Lifebook P7120.
Regards,
DigbyT
--
Digby R. S. Tarvin digbyt(at)digbyt.com
http://www.digbyt.com
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
