A howl on the wind…

Ordered a new computer Friday, received it Monday.

Comparison

What I look for…

As is probably evident from the hardware list, I value silence. I do not, however, care what the cabinet looks like as long as it has USB plugs in the front. The computer goes under the table, so I won’t be looking at it enough to care whether it is black, white, or purple.

The 320GB harddrives are Seagate Barracuda 7900.10’s, which are nicely silent. I wanted to move both drives over, but hadn’t foreseen that the DVD drive would take up an IDE slot (duh me), so I’ll have to get an external enclosure for one of them. Suits me fine as I also wanted a nice big portable drive anyways.

Relocating Windows XP via Ubuntu

As my old computer’s DVD burner was rather dead, I wanted to use the new machine to burn a Windows XP SP3 (sic) disc I had slipstreamed with latest updates and various drivers. Luckily I had Ubuntu 7.10 on dual-boot from the 320GB PATA drive, which had no problems with any of the hardware changes. So I used Ubuntu to dd the entire harddrive over to the 500GB SATA drive, basically relocating existing installations so that when the system boots from the SATA drive, partitions won’t complain about being re-enumerated. This worked without any problems and I could boot Ubuntu from the SATA drive after some /etc/fstab and /boot/grub/menu.lst edits.

I figured that my old Windows installation would blue-screen or refuse to boot since all the hardware had changed and I had not made a clean hardware profile for it. I also recalled something about Windows needing a clean reinstall to properly detect and work with multiple processors if it had been installed with only a single CPU. So I didn’t even consider booting that installation. If only I had…would have saved me a few hours.

I installed the fresh Windows XP SP3 (sic) onto a new partition and started filling it up with drivers and tools. That turned out to be terribly buggy: Sound would not work at all and applications would randomly crash. After a few hours of grumbling, I decided to see what would happen if I booted my existing Windows XP SP2.

To my amazement, it worked. No blue screen, no warnings nor errors. It simply ignored hardware it couldn’t find and happily started detecting all the new hardware, so I installed the drivers and everything just worked. Windows XP SP2 successfully relocated to entirely different hardware (even from PATA to SATA drive), and it just worked. No need to reinstall any of my applications, and it has full use of both cores in the CPU.

I must grudingly admit I had not expected that at all. I thought only Linux/BSD/MacOSX were hardware agnostic to such a degree. But I cannot complain…saves me weeks of fiddling with installations and preferences.

Here’s a little rant about a sin of which Google, Microsoft, and MySpace, and many other big names are all guilty: Automatic language detection based on anything except the actual Accept-Language header the browser sends. They instead use IP geolocation or try to guess from you picking the country you are from.

I am Danish, I live in Denmark, and my gateway is indeed in Denmark as well. However, I prefer to read English, and thus I have set my browser’s Accept-Language header to “en, en-gb;q=0.7, en-us;q=0.3“. Problem is that none of the big sites seem to respect that. Google automatically redirects me to the Danish google.dk (which I at least can force back to google.com). Microsoft sites cannot be changed without manually editing the URL to append “mkt=en-us” to the query string.

I could understand geolocation if the browser was not sending a valid Accept-Language, or to supplement and show local news or events, but please honor the nice standard header in all other cases…

Veering from the online track, in the offline world a similar annoyance struck me with Fedora‘s language and locale settings: It seems impossible to have the full Danish keyboard layout with dead keys, while also having all applications, menus, and terminal messages be English, and have the default encoding everywhere be Unicode (UTF-8). If anyone knows how to overcome that particular issue, I’d be happy to hear…

I’ve finally had enough of looking at Bitstream Vera, so I uninstalled it. I had previously merely changed the system default to Tahoma via the MS TTF core fonts, which helped immensely, but then today I stumbled upon the Try KDE page. For some reason, they have hard-coded their site to use Bitstream Vera. I suppose it makes KDE users feel at home.

I recognize that it is a free full-featured font, and as such it is nearly their only choice for distributions, but I just can’t stand it any longer. I find it to severely lack polish; some parts of letters are visibly thicker than other, and in odd places. The second font in the KDE.org list of acceptable fonts is Lucida Grande, so I tested out how it would look in that…

Comparison with Lucida Grande

• The Try KDE page in Bitstream Vera Sans (png)
• The Try KDE page in Lucida Grande (png)
• Single word: , Bitstream Vera Sans.
• Single word: , Lucida Grande
• Single word: , edited from Bitstream Vera Sans to show how it needs to be polished. All I did was remove the extra pixels in the letters S, c, e, and o. But even so, e still looks off-balance.

Should be noted that I do not use font anti-aliasing as I find it makes all text on the screen very blurry, plus I can clearly see the sub-pixel adjustments making all letters have hazy colored outlines. Doesn’t matter what engine is used (even MS’s ClearType), it only makes reading on the screen impossible. I did test whether anti-aliasing would help with Bitstream Vera’s problem, but no, it only made it much worse.

While the Apache HTTP Server is a wonderful piece of software, it sadly does not have any built-in way of running vhosts as the user they represent, which has been the headache of many an admin. Either have to fiddle with groups, running multiple daemons, or something worse. In most cases, it led to users being able to access eachother’s files. There was an experimental multi-processing module called perchild designed to combat this, but it ultimately never made it into the 2.2 branch. Speak no evil of the dead, and so forth.

If one is able to apply and maybe edit a patch, one can achieve near-perfect user seperation in a single Apache process, though.

The two projects, Telana peruser by Sean Gabriel Heacock and the ITK mpm by Steinar H. Gunderson, provide such patches. Each have their own gotchas and configuration, but once you get either of them working you’ll never want to go back to running Apache as a single global user. Why neither of these have been included upstream or in major distributions, one can but wonder…

A long time ago when I started wanting good incremental backups for my data, I looked around the net and found many good tools, but none that did exactly what I wanted. Either the process to restore from the backup was slow, or they didn’t support SSH, or any number of other small annoyances.

So I wrote my own wrapper around rsync to do what I wanted: incremental.sh. And now I’m releasing that for anyone else to use. Pretty much as long as the source and target machine can run rsync, they can use the wrapper. Instructions are in the script itself.

It makes use of rsync’s –link-dest option to minimize space usage for increments by hard linking unchanged files to the previous increment’s version. I run it nightly, and have between 100x to 1000x speed improvements over doing a full backup, plus even though I keep several days’ worth of changes, the used space overhead is limited to only what has changed between current and oldest version.

Another plus is how restoring from any of the increments is as easy as copying the file out of that folder. And a full restore from that point in time is an rsync away. I have done several full restores from backups made with this script, and I so far haven’t found a better format for backups.

One minus is that currently this is only a pull script. I have no push version, nor need of one, as I prefer running a read-only rsync daemon firewalled down to only the IPs that will be pulling data from it.

URLs

• http://svn.projectjj.com/tdtools/backup/incremental.sh