The hard drive of BOINC zombie #4 seems shot. It’s making weird ticking noises, and when I rebooted, (after unplugging it) the filesystem was so shot it couldn’t recover from the damage to the point there was no hostname – it showed up as (none) – and I couldn’t even log in. I couldn’t even reboot with ctrl-alt-del! I was figuring I’d reinstall Debian, but given how awful the hard drive sounds I figured it wouldn’t even be worth the effort. Then I realized that since it seemed all the other bits and pieces of the machine seemed to be in working order, I could run it as a live system of a CD! As luck would have it, Debian has official live CDs. So now I think I’ll wipe the drive, add it to the pile of drives to be recycled, and start running zombie #4 off a live system. It might not be as snappy, but the entire CD image is only 106MB!
Author: Steve Dougherty
New Steam Interface!
Valve gave the Steam interface an overhaul! It looks awesome! Fire up Steam and give it a look!
Compiling Gfire
Gfire is an extension for Pidgin that allows for Xfire support. I compiled mine from source, partly because I can, partly because I was too lazy to google for a .deb, and partly because there’s no official Hardy AMD64 .deb availible on the above snapshot page. Here’s what I had to do to compile:
- Download and extract source archive (should be in a directory path without spaces)
- Install pidgin-dev and imagemagick
- From the source directory, run ./configure –prefix=/usr
- Run make
- Run sudo make install
Done! You don’t even have to restart Pidgin, just add another account and it’ll be in the dropdown protocol list!
Wow!
I’ve only tested this is Hardy Heron at the moment, but it appears that, given the proper repositories, it is now possible to install applications from the browser! It’s not only an idea, it’s in practice! This’ll let tutorials consist at least partially of application installation links! ^^
Super Cache!
If the site seems maybe a bit faster, I bet it’s because of Super Cache! This plugin will store pages as HTML and serve them up instead of more intensive PHP pages with oodles of MySQL quries. It was a bit of a pain to get working. First I had to enable mod_rewrite, and then the readme told me I had to delete the configuration files of another, less extreme caching plugin that it uses to supplement itself. It then complained it couldn’t find the file. Oh well. Given write permissions to various folders and files, it was largely able to fix its own problems. You can tell if a page has been served to you by super cache by checking the source code of the page at the bottom for a comment saying as much. This means my server will probably be even LESS occupied. I’d like to run Seti@Home on the server, too, but I’ve tried it before and it loses some responsiveness and becomes horribly unstable. I suspect it’s probably overheating, which will hopefully be solved once and for all when we get our house wired with some Cat 6 and stick all but the alarm computer and the desktops in the basement. I imagine computers are quite fond of a nice 57 degrees. (Fahrenheit, I don’t want to melt them. :p)
Brasero
You know Brasero, the new (and very shiny) CD burning suite in Hardy Heron? I have had two problems with it thus far in the course of burning the RC ISO from my beta installation. My root partition for the beta only had around 550 MB left – not enough to mirror the ISO in a temporary directory. It failed with a vague error message:
Error while burning:
the selected location does not have enough free space to store the disk image (692 MiB needed).
Yes, the error message really was that poorly written. I tried another CD, thinking (incorrectly) that there might be something wrong with the CD itself, but I noticed that Brasero itself reported the CD with 702 MiB or so of space, and recordable. I poked around, and under the properties button next to the drive selection, I found a temporary directory with, you guessed it, around 550 MB free space. I told it to use my other ext3 partition, and it burned the disk happily. It then ejected the disk, then requested I put it back in for a read verify check. I did so, but Nautilus popped up first, showing me the contents of the disk, and there was also a package manager asking since there were packages on the disk if I wanted it to open. I said no. Brasero complained that the drive was in use. So much for read verify. I’ll see if I can post my errors somewhere where they can be addressed. Maybe they’re already fixed in the RC…
EDIT: Yup, it seems to be fixed now. I got a read verify to run.
EDIT2: It helps to disable the automatic display of the contents of new media in Nautilus’ preferences. To disable everything Nautilus does to the disk, you’d probably have to disable it popping anything up when a disk has software on it, too.
More Cats!
Malasysia
How many of you knew where Malaysia is? I didn’t either. It’s here. It’s a rather small country in southeast Asia. They’re getting a huge rollout of 224 megabits per second (direction not specified) broadband over power (Internet over power lines – plug an adapter into a wall socket and off you go) for $1.58 per month. FiOS is about the fastest thing available at an average consumer level, (and it’s nowhere near national rollout at this point) and it costs $43-$50 per month, offering up to 50 megabits per second downstream and up to 20 megabits per second upstream. If they can do it, why can’t we?
I’ve read about this huge island of plastic in the Pacific Ocean, but I’ve never seen it. I looked on Google Earth – it isn’t there. It makes sense to me why Google wouldn’t take extensive pictures of the ocean – in the most part it’d be a waste of satellite power. I recently found a series in which some people get a boat and go out to this patch of plastic. It’s thought-provoking. You can watch it here.
On the subject of satellites: space is full of them. Not just operational satellites, but shards of them. Here’s a picture.
Uninterruptable Power Supply, AKA Overpriced Surge Protector
At around 2 AM Wednesday morning, the power went out, and my server UPS informed me with earsplitting screeches. The logical thing to do would have been to log on the server and shut it down gracefully, but I wasn’t thinking straight, seeing as I had been awoken just then by an alarm. I unplugged the UPS, but of course it went on making its horrible, piercing tone as there was no power. I then had to hold down and release the UPS’s button, and it stopped. I then spent a few minutes bringing the machines back up to speed. I ran down to see if my rig’s UPS had held out for a few minutes, but it hadn’t. Oh well. I’m beginning to wonder if UPSes are worth it.
BOINC Zombies!
I’m done! I set up all four new BOINC zombies, and they’re all crunching away at seti@home workunits at this very moment! It’s been fun.
When we got the computers here, the cats gave them a rather thorough check:
I took off the faceplates in the hope it would improve airflow. It took me a while the first time to realize how to do it. The next few times I improved greatly. There are a total of 6 tabs holding the faceplate on. In my experience, starting on the left and going towards the right is the best. First you have to click out the tab over by the CD drive, then the one to the right of the floppy/hard drive enclosure, then the one next to the power button. Then you jiggle the bottom three catches loose and pull it off. I did it by pulling down a bit, then out. I didn’t get amazingly fast at doing it, but it works.
These were unaltered school machines. I was going to image the quality software that came preinstalled, but ran into a snag getting G4L to zero-out (not image) the free space, so gave up. It wasn’t worth the effort. Here you can see why:
There were some problems. The first machine I started setting up wouldn’t boot from the CD. I got it working by reading the manual, (*gasp*) and finding the jumper to reset the BIOS. I really liked the jumpers – they weren’t the little connector pins, but actual switches that were very well labeled. I was very impressed with the engineering on these things. The solution was to turn on jumper 2 on switch 1. I did this, and booted up. The settings were defaulted, then I turned it off and turned the jumper back off. It acted a bit weird when I turned it on: it disabled the keyboard “for security” for a few seconds before letting me hit F1 to get into BIOS, which is what I had been trying the whole time. I got in there, and it prompted me to tell it to autoconfigure itself, (the only option, why didn’t it just do it?) which seemed nice. I’m not sure if that was just a fancy way of saying it was loading the default settings. Once in, the BIOS was very nice. It listed the version numbers of just about everything, up to the point that it listed the MAC addresses of the network card and whatnot. Once in, it was easy to have it check the floppy first, then CD, then hard drive. Problem solved.
I also had some problems with the machines missing feet, which was a problem as I was going to stack them all up and didn’t want the stack to be unstable. I was going to cut new feet out of wood, which I may yet do, but I figured it wasn’t that big a deal and didn’t bother.
I installed Debian Etch on all the machines, and the installer was quite shiny, and the systems were very snappy once installed. These are running 728MHz “Copperfield” Pentium 3s, with anywhere from around 128 to 576MB of RAM. When I had a fresh system installed, I would log in, su, (switch to root) then apt-get update and apt-get upgrade. (nothing needed every time, presumably because it had gotten the latest from the interwebs when it installed) I then ran apt-get install sudo openssh-server boinc-app-seti. I added myself to /etc/sudoers so that I could run commands as root in my user account with sudo, then disabled root ssh login in /etc/ssh/sshd_config. (Force of habit, I guess. These aren’t open to the net.) I then figured out what worked for me to get seti@home running, thanks mostly to this guide. Here’s what I did:
$ boinc_cmd –project_attach http://setiathome.berkeley.edu/ <auth key>
$ boinc_cmd –project http://setiathome.berkeley.edu/ update
$ boinc_cmd –set_run_mode always
$ boinc_cmd –set_network_mode always
$ boinc_cmd –project http://setiathome.berkeley.edu/ resume
The auth key is emailed to you, and you can request it emailed from the seti@home site. For Rosetta @ Home, and possibly other sites that don’t email you your auth key, you may need to go through these steps:
$ boinc_cmd –lookup_account http://setiathome.berkeley.edu/ <email> <pass>
$ boinc_cmd –lookup_account_poll
I’m told this is made easier in versions after 5.8.15.
I also figured out how to set the IP to static more reliably. I think every time I’ve done it in the past, it’s been finicky and required some effort. I could get it working, but I was never really sure how I did it. Now I know. I believe every time I had tried to do it in the past, I had an active SSH session open with another machine as a guide. Having a guide is fine, but having the open session apparently made everything grumpy. Now I just run cat /etc/network/interfaces in SSH and log off, keeping the screen up. I then use the other machine’s config as a guide, making changes as needed, then run sudo /etc/init.d/networking restart . This works. The problem had always been having an open SSH connection. Hooray learning! Here’s an example:
#you might need
allow-hotplug eth0
#then
auto eth0
iface eth0 inet static
address 192.168.1.200
netmask 255.255.255.0
network 192.168.1.0
broadcast 192.168.1.255
gateway 192.168.1.1
I then realized that I wanted the clock to be sane. I installed ntpdate, ran it once pointed at ntp1.sbcglobal.net, then uninstalled it. I then used /sbin/hwclock –systohc –utc to set the hardware clock to that time. (As found here.)
When I was almost done, my cat saw fit to steal my chair:
At which point I decided to take a break. The result?
