🎉 Celebrating 25 Years of GameDev.net! 🎉

Not many can claim 25 years on the Internet! Join us in celebrating this milestone. Learn more about our history, and thank you for being a part of our community!

Clustering OSes and old old old old hardware

Started by
-1 comments, last by nerd_boy 17 years, 3 months ago
(Note: I may have posted some of this some time ago in regards to another question. I couldn't find it and I'm pretty sure it wasn't the same question. If so, sorry.) As an independent study with a few other CS majors, we are supposed to work on an implementing N+K Queens using Dancing Links on a Beowulf Cluster. While we were working on learning about Dancing Links, we were supposed to get the cluster up and running, both in tandem. However, the only machines that we currently have to work with are really old(maybe 4-6 years?) machines. They were used about 4-6 years ago for another cluster that nobody ever used at the university(wasn't really public knowledge, granted; hidden away on a house slightly off campus). The people who worked on it in the first place didn't do much else with it for awhile and it fell into disrepair. Only one person is still considered to "work" on it, but he doesn't and hardly stops by when we've set scheduled meetings, etc. Anyway, the hardware, as I've stated, is extremely old. It kept throwing kernel panics whenever we tried to boot it originally(CentOS 2). So we attempted to install a different operating system, including reinstalling CentOS 2. This still didn't work. The guy who "worked" on it said it was probably the harddrive. (I'm not really a hardware person, but I don't really think this was it.) We still continued to get errors for most OSes, and the ones that we didn't get them on were ones that required a CD-ROM drive on each node(like BCCD), which we didn't have. After a bunch of nit-picking and selective package installs, we finally got CentOS 2 reinstalled on the machine(yes, there were kernel panics during installs as well). However, it still continues to give the occasional kernel panic. In addition to this, the people before used Warewulf to spread the OS across the child nodes since all child nodes were composed of were a basics like RAM, a processor and four network drives. They had no harddrives, etc. However, the only version of Warewulf that would work with the machines and CentOS 2 was 2.1. I have very little experience with networking stuff, and this required that I setup dhcpd with Warewulf in some manner. I've tried to find tutorials on this in regards to 2.1 since, to my understanding, there have been quite a few changes in the recent versions, at least the instructions for them didn't seem to work with the older versin. Essentially, I would be very very grateful if someone could point me towards some resources for setting up Warewulf on multiple nodes with dhcpd. Or if anyone knows of another way to push the OS onto the other child nodes from the head node with one CD-drive, I would also be very grateful, even if this means using a different flavor of Linux. Thanks!

This topic is closed to new replies.

Advertisement