January 11, 2006: This update is long overdue. For the 2004-2005 school year, the cluster was mostly shut down; only master was kept on for purposes of serving various things. As of the end of my senior year (2004-2005), the cluster was officially shut down. It is unlikely that it will ever be booted again as a complete cluster, unless someone from the school decides to take it over. Most likely it will simply be shipped off to a recycling center. The cluster is dead--long live the cluster!
June 7, 2004: I have had to shut down node04 due to what sounds like a bad fan. I do not know if I will be able to fix it before school gets out for the summer.
June 2, 2004: The school year being almost over, I have to prepare the cluster for the summer. Mr. Nielsen, my mentor in this project, is retiring, so I do not know if the new teacher will want it here or not. For the summer, I have removed all the monitors and keyboards except those belonging to master, then moved all the boxes toward the wall. There is now a row of boxes, with master and its peripherals at the end. I also cleaned up the mess under the counter, arranging the unused computers and making changes with power cords. While doing that I accidentally unplugged a couple of nodes, but there was no damage caused. I am a bit concerned over how well the cluster will work over the summer, as master seems to mysteriously lock up and require a hard reset every few days. Since master has the only connection to the Internet, and I also start all the SETI@home processes from it, this is a bit of a concern. I will be trying to remedy this problem before school gets out.
April 23, 2004: The cluster could probably be officially classified as final now, I have not made any major additions/upgrades for quite some time now. I am experimenting a bit with network services such as Apache and Shoutcast, but other than that I am just running SETI@home. Like I said here, if somebody wanted to send me some really cool new hardware I would be glad to put it in, but the cluster is otherwise finished expanding
March 23, 2004: I recently upgraded the master node to Debian unstable, the old software in stable was really starting to bug me. When I finally got everything downloaded and installed, I found that openMosix no longer worked. The fix for that turned out to be rather simple, I just reinstalled my custom kernel. By the way, my SETI@home account for the cluster is here, as you can see it is coming along rather nicely.
January 23, 2004: Okay, seti@home runs fine, but I would like to run something designed for a cluster like mine, so if anyone has suggestions please send them to me at di gi@nu wen.net (remove the spaces to send email).
January 21, 2004: After a bit of a hiatus during which I just tweaked some settings, I have actually gotten back to accomplishing things. The school has just had its network relaid, so now I have a net connection for master. I downloaded seti@home, and sure enough, it does migrate, but only after I have started multiple instances of it. Still, that is pretty cool. Also got sound to work on node00, it has a built-in speaker, so I just use that and the CD-ROM drive.
December 8, 2003: I just got a seventh node up the other day. I had thought there were some compatibility issues with the SCSI controller and the kernel, because the install CD would just stop when it got to the SCSI controller, but it turns out that it just needed a minute or so to detect and set up the controller. I still don't have a connection to the school network, I don't know when that will come in. I had used my last power cable on the seventh node, but I went to the Moses Lake Goodwill store and bought a bunch more for 99 cents each, so I'll be good for a while now.
November 10, 2003: Added another node, I had installed Debian on it back on Friday, so it was quick. I have some kind of cold or other sickness, so I didn't really get much else done. Hardware work with a head cold is not at all fun, so I did a little tweaking, tried to get email working among the nodes. LAN work is new to me, all I had done before was over modems. (And a two-box crossover connection, but that doesn't really count.)
November 5, 2003: Today I spent the period configuring another Pentium box as a node, it already had Debian installed so it didn't take very long at all. I had a little trouble getting the networking set up, but other than that everything went smoothly. Following the naming scheme, I gave it the hostname "node03". Now I'm going to have to wait a while until I can get some more patch cables.
November 1, 2003: Yesterday I got master configured as a node on the cluster, then started the stress tests on the cluster. As far as I could tell everything was working as it should have,
October 31, 2003: I have finally gotten two of the machines to run an openMosix kernel and actually see each other as nodes. "master", on the other hand, got frotzed somehow to the point where I just couldn't get openMosix running on it. I have reinstalled Debian and will try today to get it set up for the cluster. I had been having problems with getting my Pentium machines to boot the new kernels, it turned out that they were being built for Pentium II or higher, so I fixed that and now I have at least one booting. All the nodes must run the same kernel, but it won't cause any problems to run a Pentium-optimized kernel on a Pentium II. When I had to leave yesterday I had a node compiling a set of tests to run on an openMosix cluster, I'm going to run them today on my 2-node cluster (it may be 3 nodes by the end of the day, if I get master set up) and see if everything is working correctly.
October 12, 2003: Yes, I know it's a Sunday, and with my project at school I can't work on it, but I'm going to give a summary of what I've done up to this date. I have 10 boxes, 4 Pentium II machines and 6 Pentium I machines. I also have a couple of old monitors that aren't really good for anything except console mode (no X). A couple old ATX keyboards, two PS/2 keyboards, and a few mice (serial and PS/2). So far I've focused on getting the fastest machine, which will serve as a master node, running correctly with Debian and two ethernet cards. There were some problems with the ethernet cards, they didn't want to work together, but I finally got it straightened out.
Maintained by John Floren, di gi@nu wen.net (remove spaces to send email).