New box

With $600 of my own money, and $250 donated by friends, I’ve build my new box. It’s pretty nice.

It starts with a Rosewill R5604-TBK case. Not the fanciest aluminum case like the cool kids have, but still pretty nice. As the picture shows, it has a hatch on the front with power switch, reset switch, audio, USB and Firewire ports. Another hatch hides the place for the floppy (which I’m not going to bother with) and up to 4 externally accessible 5.25″ drives – I have one IDE 16x DVD burner that I’ve borrowed from the old machine and that will probably be it for 5.25″ drives. The case is designed to not require tools, and there was a plastic box in the internal drive space with rails for the drives and other hardware. It also uses the new very large and slow and quiet case fans, and those fans are backed by cleanable filters. Another nifty feature of this case is blowhole on the side, which has an extensible tube that lowers down over the processor’s heatsink and fan so that the cpu ejects its heat outside the case directly.

To that case, I added a 500W power supply, Intel DQ965GFEKR motherboard, Intel Core2 Duo E6320 processor, 2Gb of DDR2-800 RAM, and 2 500Gb SATA-II drives.

Old Machine New Machine
Processor 2 32 bit Athlon MP1800+ 512KB cache 1 64 bit dual core Intel Core2 Duo E6320 1.82GHz 4MB cache
RAM 2 512Kb PC133 ECC RAM 2 1Gb DDR2-800 RAM
Disks 1x160Gb, 1x200Gb, 1x250Gb, all IDE 2x500Gb SATA-II
Ports 2xUSB 1.1, Serial, Parallel, 3xIDE 6xUSB 2.0, 2 Firewire, 10/100/1000 Ethernet, Serial, Parallel, 6xSATA-II, Audio, Video
Fans 3 8mm case fans, 2 8mm heat sink fans, 2 8mm power supply fans 2 12mm case fans, 1 7mm heat sink fan, 1 12mm power supply fan
Reliability Crashes twice a day Ran memtest for 12 hours straight with no problems

It’s all pretty sweetly put together. There are only a couple of strange things:

  • The case has a speaker connection with nothing to connect it to on the motherboard. Which is strange because the motherboard manual has information about BIOS beep codes – a little hard to hear beeps if you don’t have a speaker!
  • The BIOS has a RAID setup, but when I booted Ubuntu, it saw it as two separate disks. And when I set up the two disks using software RAID, things got very confused.

So now I have the machine set up, it’s time to get the software installed. I decided that rather than try to copy everything over from the old box, I’d set the new box up as a brand new box, and copy individual files and accounts over as necessary. The old box is a lineal descendant of others and has been upgraded from early versions of RedHat, and hasn’t had a clear wipe and re-install since the late 1990s. As well as the cruft left over from all that, I’m also getting severely pissed off at the lack of quality of Fedora 7. So I decided to give Ubuntu a try, since that seems to be the current reining champion of distributions. It’s also based on Debian, and I’ve been using Debian on my servers and I’m pretty pleased with how good it is at solving dependencies and even more important, removing stuff that isn’t needed any more.

First problem, when I booted the Ubuntu CD the monitor went into “INPUT SIGNAL OUT OF RANGE”. Even trying to force the “safe video” mode didn’t help. I burned about 5 different CDs trying to find a combination of i386 or AMD64 versus Ubuntu/Kubuntu/Xubuntu versus normal or alternative that worked. I ended up installing using the text-only mode, which looks a lot like the Debian installer.

I set it up with a 3 partitions on each disk, a 2Gb, a 1Gb, and one for the rest. The 2Gb ones were /boot and /, and the 1Gb ones were swap. The “rest” partitions became the two halves of a RAID 1 (mirror). Then that were made the LVM (Logical Volume Manager) VG (Volume Group). And then I carved partitions out of that – I’m the sort of guy who likes having lots of different partitions, and by using LVM I can fix it if I got the sizes wrong.

The basics are now installed and I’m trying to get the various services that I have on the old box up and running on the new box.

I eventually got X running. The problem appeared to be that no matter what you put in the xorg.conf, the X server would query the monitor, and the monitor would lie and say it could do a vertical refresh rate that it couldn’t actually handle. I couldn’t find anything in the xorg.conf file to fix this, but I discovered that if I changed the boot options to include “nodcc”, it worked. Have I mentioned recently how anybody who says Linux is as good as Mac OS X is a fucking idiot?

5 thoughts on “New box”

  1. The BIOS has a RAID setup,

    Many claim to, but AIUI, it’s a cheap hack that expects to be taken over from by special Windows drivers once the OS starts. There’s not really much point if the OS has decent soft-RAID of its own.

    (Bah. I was going to comment with OpenID, but LJ thinks I’m not logged in, except that it also thinks I am logged in.)

  2. So in parallel to the “WinModem”, we should call this “WinRAID”?

    Sorry about the OpenID problem – I use the SpamKarma2 plugin to prevent comment spam, and it and the OpenID plugin don’t get along. As far as I can tell, whichever one runs first sometimes swallows the hidden form variables that the other relies on. Some people can post using OpenID, and others can’t, and I can’t figure any common difference between those who can’t and those who can.

  3. At least in my case, it’s not you — it seems to be Opera.

    As far as I can tell, Opera is withholding cookies from any page it’s been redirected to.

    I discovered this by accident yesterday with Google, when I was finding export.bml; at the time I thought it was just Referer madness, but with two instances on my hands I was curious and looked a little closer. (Google was a little confusing, because it looks like they’re not using a redirector, but in fact their links have JavaScript mousedown handlers that change the href attribute as you’re clicking on it.)

    Now, to try to find out *why* it’s doing that….

  4. Aha — it’s the “accept cookies only from sites I visit” option. Apparently this also affects how already-accepted cookies are returned, and applies to redirection targets as well as images and such.

    Well, that answers that, I guess.

  5. I’m not sure if it’s relevant to your cookie problem, Jed, but as far as I can tell, in order to do OpenID logins, the site does a redirection to your OpenID server (such as LiveJournal) with some form variables set, and the OpenID server is supposed to set some more form variables and do a redirect back to the originating site. Since the OpenID server is on livejournal.com, it should be able to read livejournal.com cookies.

Comments are closed.