Hi. I'm trying to set up some automatic backup scripts and I am hoping
that some of the more seasoned Unix admins out there can share some wisdom
with me. I will summarize the responses in detail.
I am running DU 3.2f, Columbia Appletalk Package, BIND, and Apache httpd
on this server. It has a 2GB hard drive, with a 63 MB (ufs) root
filesystem, and a 866 MB (ufs) /usr filesystem, plus a Digital TLZ09 tape
drive (up to 4GB). This is strictly a web / DNS server, so no users ever
log in except myself.
My specific questions are as follows:
1) Am I asking for trouble if I do backups of the /usr filesystem while
it's still at run level 3? What about the root filesystem? Should I
make my script shut it down to single user mode, start my backup from
a script in /sbin/rc0.d, and then reboot? Maximimum uptime is very
important to me, but data integrity is more important.
2) I definitely want to backup the /usr filesystem every night. Since my
root filesystem is so small, I'd like to just go ahead and back it up
every night also. According to my interpretation of the man page, the
following commands should backup both filesystems to the same tape...
dump -0uN -d 61000 -s 76586 / ; # The 'N' flag means no rewind
dump -0u -d 61000 -s 76586 /usr
...but when I do a restore -i, it only lists files in the /
filesystem. I can't find any way to get it to skip to the next volume
on the tape, or to even acknowledge that there is another volume on
the tape.
So, what do I have to do to backup multiple filesystems to the same
tape?
3) Finally, does the following dump command look OK?
dump -0u -d 61000 -s 76586 /usr
The dump seems to go OK, and I've been able to restore single files
successfully, but when restoring I sometimes get a message about 'Read
error while skipping inode xxxx'. I'm scared that one day I will lose
my whole drive and find that my backups are useless.
A *million* thanks for any help you can offer.
Derrick Miller
Herald-Mail Co.
Hagerstown, MD
Received on Tue Aug 27 1996 - 03:30:50 NZST