I’ve been tracking the Debian kFreeBSD port for a while now, and it’s really starting to mature.

The latest “Squeeze” release of Debian aims to have a reasonably stable release of Debian running on top of the FreeBSD kernel.

I’ve done an install from the recent d-i.debian.org installer and things are really looking good, aside from a few niggly issues.

Why is this good ?

Well, an alternative kernel to run the Debian userspace atop of is never a bad thing in my mind. In the same way the Nexenta has utilised the OpenSolaris kernel to build a Debian userspace atop.

With a FreeBSD kernel and Debian userspace, I get all the tools that I need and some kernel specific features, such as “pf”, “zfs” and a number of things that the mainline Linux kernel doesn’t have.

Why is this bad ?

Well, as per the DFSG (Debian Free Software Guidelines) some code that isn’t freely licensed is being hacked out of the kFreeBSD kernel. Importantly for example Atheros WiFi device support and who knows what else.

This is not much different from Debian Linux releases (for example bnx2 ugliness) but it does kind of remove some of the benefits of a FreeBSD kernel.

In the end…

Of course, I’m probably just going to end up hacking the kernel builds to include all the proprietary firmware, because as a developer, sysadmin and user I don’t really give a shit about DFSG.

But that’s the beauty behind DFSG. It’s a “guideline”. I can choose to ignore it if I have enough gusto and know-how to do it.

Of course, this is part of what Mark Shuttleworth and Ubuntu’s entire selling point is.

Hmm, I spot a business model here. Kind of a must-make-funky-presentation-to-daft-venture-capitalists moment here. 😉

After having “debootstrap”-ed my home firewall three times in the last week due to harddrive failure, all I can really say is:

1.) Reliable disk space is worth every cent
2.) Knowing how to recover from a hdd going click-click-click is a boon.
3.) Andrew Alston and the TENET team deserve props to the 1+E06 for maintaining debian.mirror.ac.za

If it wasn’t for my ability to bootstrap a system quickly via HTTP, I’d have really been up the proverbial shit creek.

CDROM media is *so* 90’s.

In case anyone really cares, the simplest way to recover from a major HDD crash on a debian system is to get the debian-live rescue CD, and bootstrap a new system like so:

mount /dev/sda1 /mnt (new partition already formatted with your choice of filesystem evil)
debootstrap --arch amd64 squeeze /mnt http://debian.mirror.ac.za/debian
mount -o bind /dev /mnt/dev
LANG=C chroot /mnt /bin/bash
echo foo >/etc/hostname
mount -t proc proc /proc
apt-get install linux-image-2.6-amd64
apt-get install grub
grub-install /dev/sda
update-grub
passwd root
apt-get install ssh
reboot


Et voila. A basic debian system. This can also be neatly used to usurp an existing (e.g. Redhat) system if you can get hold of all the debootstrap perl code and dependancies.

Having a broadband connection available is obviously extremely useful.

I managed to do all of this via my Windows machine after having installed a tiny http proxy. Thus I could simply use

http_proxy=http://192.168.0.8:3128/ apt-get install or
http_proxy=http://192.168.0.8:3128/ debootstrap squeeze /mnt http://debian.mirror.ac.za/


I’ve used this technique before as well with SSH tunnels to a target system that didn’t have internet access, whereas my local desktop did.

Cheers TENET.