Date: Tue, 20 Sep 2005 22:27:05 -0400 (EDT) From: "Brian A. Seklecki" <lavalamp@spiritual-machines.org> To: freebsd-questions@freebsd.org Cc: bseklecki@collaborativefusion.com Subject: 220.backup-bsdlabels (WAS: Re: periodic(8) / daily bsdlabel / fdisk / softraid*) Message-ID: <20050920214827.M20769@arbitor.digitalfreaks.org> In-Reply-To: <20050920200555.B20769@arbitor.digitalfreaks.org> References: <20050920200555.B20769@arbitor.digitalfreaks.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Here's a basic script to accomplish this. No support for checking
geom/gmirror/vinum configurations. RAIDFrame's raidctl(8) has a nice
"-G" flag:
"-G dev Generate the configuration of the RAIDframe device in a
format suitable for use with the -c or -C options."
Maybe someone more intimate with these 3 RAID APIs on FBSD can offer some
insight.
Obviously, this script will need to be re-written to do sanity checks
especially with secure file operations. NetBSD has a nice function in
/etc/security called migrate_file() {...} for safely rotating. And lines
42 and 44 are just ugly.
Also, do non-i386/amd64 platforms support fdisk(8)? I know NetBSD uses
sunlabel(8)+disklabel(8) on Sparc64?
But this does what I need it to do. Given a solid off-site tape backup of
/var, I can recover /var/backup/(fdisk|disklabel) to a temp machine, and
recovery my slices.
In the event of a complete system loss, I can use fdisk(8) output to
verify that the number of sectors on my hardward or software RAID
meta-device match as they were previously on the newly created RAID.
PR is misc/86388!
http://digitalfreaks.org/~lavalamp/220.backup-bsdlabels
Thanks all!
~BAS
---------
#!/bin/sh
#
# $FreeBSD: src/etc/periodic/daily/220.backup-bsdlablels******
#
# If there is a global system configuration file, suck it in.
#
if [ -r /etc/defaults/periodic.conf ]
then
. /etc/defaults/periodic.conf
source_periodic_confs
fi
case "$daily_backup_bsdlabels_enable" in
[Yy][Ee][Ss])
bak=/var/backups
disks=`sysctl -n kern.disks`
if [ -z "$disks" ]; then
echo '$daily_backup_disklabels_enable" is set but no disk probed
by kernel.' \
"perhaps NFS diskless client."
rc = 2
else
for i in $disks; do
# first order of business is to check for an existing
backup-backup
if [ -f $bak/fdisk.$i.bak ] ; then
rc=1
echo "rotating $bak/fdisk.$i.bak"
cp -p $bak/fdisk.$i.bak $bak/fdisk.$i.bak2 || rc=3
fi
echo "backing up fdisk for $i"
fdisk $i > "$bak/fdisk.$i.bak" 2>/dev/null || rc=3
# again exept now we have to get a list of
patitions/slices
# sparc64 can have...9 hopefully slices on a sunlabel?
part_slices=$(echo /dev/${i}s[0-9])
for j in $(echo "$part_slices" | sed 's/\/dev\///'); do
if [ -f $bak/disklabel.${j}.bak ] ; then
rc=1
echo "rotating $bak/disklabel.${j}.bak"
cp -p $bak/disklabel.${j}.bak
$bak/disklabel.${j}.bak2 || rc=3
fi
echo "backing up disklabel for ${j}"
disklabel /dev/${j} > "$bak/disklabel.${j}.bak"
2>/dev/null || rc=3
done
done
fi;;
*) rc=0;;
esac
On Tue, 20 Sep 2005, Brian A. Seklecki wrote:
>
> All:
>
> I just realized that the stock perdiodic scripts dont backup fdisk/disklabel
> output. I'm taken back a bit; NetBSD and OpenBSD have always done this
> (archive to /var/backup). We backup the password and group files, but not
> system info.
>
> Obviously, RAID can mitigate the need for this, but imagine a DRP scenario
> where you have to recover a host to an offsite warm/cold failover facility
> from off-site tape backups. You can get your vendor to provide identical
> hardware, but without disk parition/slice info, how are you going to recover
> your disks?
>
> It's bad enough we don't have a unified way to talk to RAID controllers
> yet...the least we can do is try.
>
> We should probably also backup gmirror/gvinum config outputs...
>
> *sigh*
>
> ~BAS
>
l8*
-lava
x.25 - minix - bitnet - plan9 - 110 bps - ASR 33 - base8
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050920214827.M20769>