Intel has put a lot of money and effort behind its open-source video drivers for Linux, which is the main reason I went with the Intel X4500MHD video option on my new laptop, instead of the faster NVIDIA option. Going with a 100% open source solution gives two main benefits: everything works perfectly out-of-the-box, and you get features like monitor hotplugging that have set to be implemented in the closed drivers from NVIDIA or AMD.

There’s just one drawback — the Intel drivers, and the Linux video stack in general, are going through some major updates right now, which results in some odd bugs. OpenGL-accelerated desktop effects seem to work perfectly, but some windowed OpenGL apps glitch parts of the display outside of their windows, and there are also some GLSL and depth-buffer bugs that cause visual issues with some apps, like the Penny Arcade Episodes.

That work is still ongoing, but with the latest alpha of Ubuntu 9.10, the results are starting to show. I installed it on to a spare partition on my laptop today and did some testing, and I’m glad to say that the above-mentioned bugs seem to have all-but-disappeared. Windowed OpenGL apps ran just fine, and the graphical corruption in Penny Arcaide Ep 1 has disappeared. The problem now is stability — X crashed a few times when launching OpenGL apps during my testing — but with any luck those bugs will be ironed out before the 9.10 release in October.

While I no longer have a Mac, I still need one from time-to-time for development and testing purposes at work. My old MacBook Pro is still there, and runs just fine in that role, but for a laugh I thought if it would be possible to run Mac OS X in a virtual machine on my new laptop using KVM, the open-source virtualisation system built in to the Linux kernel. The answer, surprisingly, was yes:

For the installation, I used iDeneb, a bootable Mac OS X installer DVD that’s pre-patched and ready to boot on most PCs. The only trick to getting it working on KVM seems to be to use an alternative kernel — the Voodoo 9.5.0 kernel worked for me — and to find a virtual network card that both KVM and OS X support. In fact, I haven’t quite figured that last one out yet, but I’m giving it another go right now.

The only real problem is the performance — it’s terrible! Dock bounces aren’t scientific measure of how long it takes to start applications, but when you get to double-digits starting Terminal.app, you know you’re in trouble. Even if I do get the networking issues sorted, I doubt it’ll be usable for anything serious.

So… everything went exactly according to plan yesterday with my hard drive juggling: the data copy finished up on Saturday morning, the system booted from the new drives after some boot loader fiddling, and the arrays rebuilt in to full redundant mode once I added the two 500GB drives from the old array.

There was just one snag — I had to sacrifice my desktop PC.

Everything was going perfectly until I took the new 1TB drives, brimming with data, and placed them in to the TV box. It was a nice system about five years ago, but neither of the SATA controllers on the motherboard would detect those shiny new drives, despite putting jumpers in place to switch the drives in to SATA 1.5 Gbit/s compatibility mode. With no other option at hand, I swapped the TV box for my desktop box and got everything working there instead.

It’s actually quite nice having that box in the lounge, now — it’s super-quiet, and it has no trouble playing HD streams. I’ve put the old TV box back together as a desktop for now, but with just 512MB of RAM it doesn’t have the grunt to run my usual selection of music apps.

One option to solve this would be to shuffle things back around and throw a PCI SATA card in the TV box to get those drives working, but in the long run, and with the crazy-cheap prices on basic components these days, it’s probably better to build either a new TV box, or a new desktop.

Things are going well with my hard drive shuffling so far — the new drives are in my desktop box and tested, I’ve created and formatted the new arrays, and I’m now copying the data across. Over my 100Mbit network it’s probably going to take until Saturday morning for everything to finish copying, but that’s no big deal, since there’s no need to take the TV box offline while I copy the data from it. Thanks to rsync, I can do another quick copy run once this bulk copy is finished to pull across any files that have changed since I started the first copy.

Unless something goes horribly wrong, I should be all-but-done by Saturday afternoon.

I finally managed to lose a drive on my TV box — one of the old IDE 250GB drives bit the dust the other day. I’d drop in a replacement, but getting an IDE drive isn’t the easiest thing to do, and the box is out of SATA ports. So, instead, I’m going to build a new array, out of two 1TB drives and the two 500GB drives from the current array.

The whole setup is a bit convoluted, but it works quite well. The current array is actually two arrays of 4x250GB drives: each uses two 250GB drives, and two 250GB partitions on a pair of 500GB drives. I’ll be using the same approach with the new setup, with two 3x500GB arrays but getting from A to B is going to take some doing:

* thoroughly test the new drive and make sure it’s working 100%
* set up the two 1TB drives on my desktop box as a degraded three-drive RAID-5 array
* set up the arrays and LVM setup on the new drives
* copy everything from the TV box to the new drives over the network (yay rsync!)
* take all the drives out of the TV box and replace them with the two new drives
* fiddle things until the TV box boots from the new drives
* throw the two 500GB drives back in and add them to the arrays and let the arrays rebuild

That should about do it