Mondo is an excellent backup tool for Linux, which can do a backup of a running system and afterward perform disaster recovery, if your system should ever crash.
It is open source and is part of the Debian repository. In order to install it just perform a

apt-get update mondo

After installing it, you can start it by typing

mondoarchive

You’ll get a simple and nice user interface (UI), which is fully text based. The latter fact has the big advantage, that you can start it remotely in a ssh session.

But..
…don’t use it under the following conditions:
I’m running Jaunty Jackalope in x86 (32bit) environment and using version 2.2.8 (at the moment the most recent one).

I’ve you navigated through the few simple and easy understandable masks, you might come up with a very annoying bug. Mondo tries to use a udev located scratch path for its temp files, and at least if your backup exceeds 1GB, what is very likely, it will fail with a ‘no space left on device’ error in /var/log/mondoarchive.log.

Your tempfs, what is nothing less, than your RAM disk will be after a while at 100% and it will getting difficult to access your system remotely, because vital system activities can’t allocate disk space of your RAM disk. All udev devices (they reside also there) are affected.

I’ve got this in the log file:

libmondo-archive.c->create_afio_files_in_background#1195:
[5582:1] - archiving 1...
afio: "//tmp.mondo.114/tmpfs/1.afio.bz2" [offset 349m+768k+0]: No
space left on device

and the gentleman in this posting writes:

The solution was already posted in the devel group. It requires
logging in via another shell and remount /dev/shm to a much large size
(in my case 3gig).

But I haven’t found anything, that worked for me. I’ve tried to symlink, but mondo simply chose every time I’ve created a symlink to somewhere else, another path.
I’ve also tried to increase the size of the ram disk in /etc/fstab onto 3GB like this:

tmpfs /lib/init/rw tmpfs nodev,nosuid,noexec,size=3072M,nr_inodes=1 0 0
udev /dev tmpfs rw,mode=755,size=3072M 0 0

and I’ve also shut down udev ( /etc/init.d/udev stop), but it failed all way long.

Workaround

The workaround is simple. Just don’t use the UI. The command line parameters looks at the first glance a bit confusing, but here‘s simple howto, and on the next page you can read how I’ve manage to get a nice backup just without a GUI.