Subject: dump and tar problems
To: None <port-pmax@netbsd.org>
From: Daniel Senderowicz <daniel@synchrods.synchrods.COM>
List: port-pmax
Date: 12/29/1999 20:37:22
Hi Fellows,

I have a running 1.4.1 system (rz0) that I want to duplicate
into rz3 to use on a different machine. So I started the install
so the partitions were set up and newfs.  After I did:

# mount /dev/rz3a /mnt
# cd /mnt
# dump 0f - / | restore rf -

with the following outcomes:

A few times it crashed, and going into reboot after writing
something like asc_... whatever,

A few times it completed but the files were full of errors.

I checked the hardware and everything is fine (change disks,
cables, cpu, etc). Then I did:

# dump 0f - / > /usr/tmp/first
# dump 0f - / > /usr/tmp/second

and happened that first != second . So thinking that dump was
broken, I tried tar:

# (cd /; tar --create --file - . --exclude usr --exclude mnt) | tar xf -

and only a couple of directories were copied, as if tar wasn't
seeing everything. So I started tar'ing by hand, and at certain
point the thing also went bezerk starting the reboot.
Needless to say, between each attempt I was doing the first phase
of the install (newfs, etc.) to make sure that I had a clean
disk for the next try.

A fellow netter told me that there was a bug in the SCSI
interface for which there is a patch. So my questions are:

1) Where can I find such a patch,
2) I never patched a kernel, can you explain to me how to do it?

Many thanks in advance.

Happy holidays!

Dan