but the fourth one stayed up

Thanks to some generous donations and my tax return, I have obtained parts and put together a new server.  This blog and a few other minor things are already running on the new machine, though as I write this, most of my other Internet services are still on the old server, saidar.  The new server is named frodo.  I was going to name it bilbo, but changed my mind after reflection.  Frodo is a more mature character than Bilbo, capable of carrying a heavier burden.

The specs of the server are as follows:

  • Old 2U rackmount case with 460W PSU
  • ASRock M3A785GXH/128M motherboard
  • AMD Phenom II X3 705e processor
  • 2x2GB DDR3-1600 RAM
  • two 1TB SATA2 drives, mirrored

Internally, the CPU is a full quad-core Phenom II, but sold as a triple-core. This is part of a line of CPU models that allow AMD to sell chips with bad components that would otherwise just get junked. Not all of the chips are actually bad, though – some of them are perfectly functional chips that have had components disabled to sell at a lower price point. Anecdotal evidence suggests that as many as 70% of them are fully functional. The BIOS on this motherboard has the ability to re-enable the disabled components.  In addition to enabling all the cores, I have also overclocked it from 2.5 to 3.0 Ghz.  It’s been running for over a week now (since 2009-Feb-09) without any issues.

On the software side, this will be the most feature-laden system yet.  Some of that will be just because everything is a newer version, but I have learned some things in the last few years about mailserver configuration that will bring a lot of new functionality to my users.  Some of it is available on the current server, but only to people with root access.  In other words, only me.  Here’s what’s installed or planned so far:

Mail Server:

  • postfix
  • dovecot (backports.org)
  • amavisd-new (backports.org)
  • spamassassin (backports.org)
  • clamav (debian-volatile)
  • razor
  • pyzor

Web-based software (php apps are from source):

  • apache
  • php
  • phpmyadmin
  • postfixadmin
  • mailzu
  • squirrelmail
  • roundcube

I had been planning on doing some full documentation on my wiki of how I’m configuring it, but in the interests of actually getting it done, that hasn’t happened.  I have put up some disjointed notes, which I will flesh out and clean up.  Unfortunately, using those notes won’t be possible for a novice.  If anyone is interested in setting something similar up, I may be able to help out.  Such help would be free of charge unless it’s for a business.

more on google’s favicon rebranding

I’ve been seeing a lot of chatter on the Internet about Google’s new favicon. Seems the general consensus is that people hate it. I don’t hate it, but as I twittered a couple of days ago, it’s messing with my mind. When I said that, I meant that as I glance at my open tabs, I see the new icon on all my searches and I can’t immediately associate what I see with Google. I wonder for a moment what I’m looking at, until it filters up into conscious thought and I can associate it properly.

I don’t think it’s a good idea for Google to go and change their branding. It’s not that the new icon is ugly or anything. The problem is that it doesn’t scream Google to the subconscious mind. The old icon did. When you have something that works, stick with it unless you find something that works better.

Here we will take a little detour, which I promise has a point. Have you ever noticed that the word “classic” is a little like the holy grail of marketing? The only reason we don’t see it on everything is that it is a double-edged sword. If you claim that your product is a classic but it actually isn’t, you run the risk of irritating your audience.

Although I don’t actually believe it to be true, sometimes I wonder if New Coke was an expensive and risky marketing strategy designed to allow Coca-Cola to call their product “classic” without worrying about the public crying foul. I was 15 when that fiasco happened in 1985, and I remember a lot of people stocking up on the old version before the new one hit stores. New Coke was quite a lot sweeter, and as I recall tasted much like Pepsi. My comment at the time was “If I want a drink that tastes like Pepsi, I’ll just buy Pepsi.”

The return of the original formula as Coca-Cola Classic was something I celebrated. The joy of having my old friend back in my life made me love it even more. The Wikipedia article I linked above suggests that it was like that for most people. Sales surged and they strongly solidified their number one position, which is what they had been trying to do when they changed the formula.

Perhaps Google is trying to reap similar rewards. If you change things in a way that results in backlash, wait for public outcry to get loud, then bring back what people loved before but change it a little bit to make it even more appealing, it can solidify your position.

mucking with my trousers

On my personal cellphone, I occasionally send text messages. I was writing one a few minutes ago to a buddy of mine because I want to verify that I can use his server for backup DNS. I use the “multiple tap” method for text entry, because I haven’t gotten the hang of T9 yet. Even so, my phone does make word recommendations when the previous word is completed. I was typing along and had entered “I am mucking with” … its suggestions for the next word were disk, the, and your. Not what I was after, but relatively reasonable. I typed in the actual word I was after, making it “I am mucking with my” … and it popped up with “Trousers” as the only suggestion.

WTF? What sort of deranged default dictionary are they using?

Later in the message, I typed in “permission to do my” and it popped up Trousers again, but this time it had learned from before and also presented dns as a choice.