Nuclex

I'm running a small home server that, amongst many other things, stores my music collection, software, virtual machines and more. As any computer user is able to testify, no matter how much hard drive space you've got on a rig, you will run out of it eventually.

So last week, I decided to replace my aging 500 GB drives with three brand new 2 TB drives (my choice fell on Western Digital's WD2000EARS drives, btw - only 5400 RPM but dirt cheap and because this is for long-term storage, all I want is space, not speed). I partitioned all three of them using the exact same commands in fdisk (launched with fdisk -c -u /dev/sd*). Before I did that, all drives appeared with identical informations when I ran fdisk -c -u -l. Now?

The first drive now has 18 heads and 6 sectors/track:

Disk /dev/sdb: 2000.4 GB, 2000398934016 bytes
18 heads, 6 sectors/track, 36176196 cylinders, total 3907029168 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0xa28d6fec

   Device Boot      Start         End      Blocks   Id  System
/dev/sdb1            2048  3906996224  1953497088+  83  Linux

The second one got 69 heads and also 6 sectors/track:

Disk /dev/sdd: 2000.4 GB, 2000398934016 bytes
69 heads, 6 sectors/track, 9437268 cylinders, total 3907029168 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x78f7deaf

   Device Boot      Start         End      Blocks   Id  System
/dev/sdd1            2048  3906996224  1953497088+  83  Linux

And the third one got 18 heads again, but only 3 sectors/track:

Disk /dev/sdc: 2000.4 GB, 2000398934016 bytes
18 heads, 3 sectors/track, 72352392 cylinders, total 3907029168 sectors
Units = sectors of 1 * 512 = 512 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0xbef707e2

   Device Boot      Start         End      Blocks   Id  System
/dev/sdc1            2048  3906996224  1953497088+  83  Linux

I know that heads and sectors on modern hard drives no longer have any relationship to the physical number of heads and sectors on the drive, but why do I get 3 different results from running the same command 3 times, on 3 identical drives?

They all have the same capacity in the end and my raid array is running fine, but this confuses the hell out of me :D

I just love listening to music in bed, right before going to sleep. You can close your eyes and focus, experiencing the music like a movie and less like the background noise music is often degraded to during the day. After moving my home office to a different floor, however, my bedroom is no longer close enough to my PC to do this (I used a rather quirky solution, using a game pad to remote-control WinAmp and later shut down my PC :D).

This put me on a quest for a replacement. Because I ripped my entire CD library as FLAC and stored it on my home server (running Gentoo Linux and serving the music files as a network share via Samba), I was looking for a player that supported FLAC and, ideally, would play music from a standard network share. This is what I ultimately decided on:

That's a Logitech Squeezebox Boom.

I admit that I half expected this category of devices to not exist at all, because a "normal" person doesn't have a home server or a PC that's always on. And when I found something I expected some half baked showpiece hardware, but was yet again positively surprised. Read on for my personal review of this brilliant device!

Last Friday (and I only now notice it was Friday the 13th :D) my screen stopped working. I dismantled it and found some bad capacitors, then decided to do a small foto story showing my attempt to get it working again: My Screen went Dark :-(.

Today the electronics components I ordered arrived and I could finally replace those capacitors I found to be broken last time.

Soldering in the new capacitors was surprisingly easy. I remember that when I did this in my childhood, I spread solder everywhere except where I wanted it to ;)

Earlier this year, one of my monitors started behaving strangely each time it was turned on the first time for the day. The image would flicker on and off, first very slow, maybe twice a second, then faster and faster still until it displayed a permanent and stable image.

Over time, things got worse. First it would take just a few seconds, then two months later, the scree would stay black for minutes before the now familiar flickering started and the display settled. This morning, the display just remained black.

Some googling revealed the likely cause: bad capacitors. Between 1999 and 2007, many electronic parts were sold with bad capacitors because, at least that's a popular story, one Taiwanese company had obtained the knowledge to build electrolytic capacitors via espionage, but the informations were incomplete and the electrolyte was missing and certain agent that prevented the hydrogen from escaping.

Whatever the reason, my TFT's production date falls into the problematic range and symptoms are similar to things other people reported. So I went ahead and tried to take a look at the thing, documenting each step with my camera.

Being an adherent of Continuous Integration, I need a build machine that runs round the clock even when my workstation is turned off. As I'm running a small home server, this wouldn't be an issue -- if it weren't for the fact that my home server runs Linux and 99% of my development happens in Windows. So I use virtualization to run a small Windows system on top of my home server.

In the past I used VMware for this job. VMware worked well for me and performance was quite good, but now that I've switched to a fully headless system, I noticed that the vmware-server package pulls in most of the X11 libraries - which I'm not particularly keen on having on my system due to their compile times.

So I went shopping for some alternatives. KVM sounded interesting (and was the leanest virtualization solution I could find), but the Gentoo Wiki stated that Windows didn't work in qemu with recent kernels, so I went looking on - and found VirtualBox. This article explains how to set up VirtualBox on a headless Gentoo system.

If you haven't heard of Continuous Integration yet, it's the practice of setting up an automated system that rebuilds projects automatically whenever someone commits a new change to your source code repository. It ensures that whatever is in your repository builds and runs: automated builds usually involve compiling, running unit tests and packaging the installer.

To do continuous integration, you need a tool that monitors your source code repository and starts the builds - a continuous integration server. My weapon of choice is TeamCity, a free CI server written in Java with first-class support for .NET and its toolchain (like NAnt, NUnit, NCover or PartCover).

TeamCity is pretty easy to deploy - the Windows package has an installer which leaves you with a fully working server after just a few clicks and even the Linux package is pretty simple to deploy: Download, unzip, run runAll.sh and you're done. To properly integrate it into a Linux server (so it will come back up after rebooting and can be reached via HTTP without having to run either Apache or TeamCity on a non-standard port), you'll need to run your own Tomcat server.

This guide will tell you how to do it!

Courier not only is an excellent mail server, it also ships with a mailing list manager that can be used to build mailing lists without relying on a third party provider (which usually has the bad habit of adding advertising text to the emails being forwarded).

Here's a small tutorial that explains how to set up a new mailing list using couriermlm.

Please excuse the current flurry of Linux articles. I'm moving servers and this is my way of writing notes to myself and possibly helping out others. Normal service will resume shortly ;-)

This article is a follow-up to my guide on Installing Courier on Gentoo. As long as you have a working Courier installation on your system, there should be no issues following this guide.

Running a mail server without some kind of spam filtering is just insane these days. SpamAssassin is a nice solution, especially if you run SpamAssassin during the SMTP transaction to reject spam while it is being uploaded to your server.

On my previous system, I had used qmail (netqmail actually, which is qmail with some patches). Qmail is moderately difficult to set up and in its 3 years lifespan on my system, it has broken down on several occasions. That's why I decided to use another mail server when I moved my domains to a different system.

Because the Courier IMAP server has never let me down before, I decided to give the Courier Mail Server a chance. Lots of people are using Courier IMAP to access their mail but Exim, Postfix or Qmail to accept incoming emails. Even the Gentoo Wiki contains various HowTos for these combinations, but not a single one for a homogenous Courier setup. After trying out Courier, I don't see why, so this is my attempt to rectify the situation (and to remember what needs to be done for the next time I'm moving my domains to another system!)

If you want to download torrents on your Linux system, there are several clients to choose from. One of the nicest and fastest clients is rTorrent. It is full-featured, supports encryption, dynamic host table exchange and achieves fantastic download speeds.

But its best feature probably is that it isn't bound to any windowing toolkit. You can install one of its GUI frontends to manage it on your fancy KDE 4 desktop machine, but you can also run it on a headless system and manage torrent from a text-only console. And if you happen to run it on a home server like me, there's wTorrent, a beaufitful AJAX-driven web frontend that allows you to manage your torrents in your browser.

Installing wTorrent isn't the easiest thing to do, so, as when I tried to get the best out of my SSD, I decided to write this small article explaining how to do it. I'm using Gentoo Linux, but it shouldn't be too hard to apply this article to another Linux distribution.