Slashdot Mirror
ClusterKnoppix
chronicon writes "Knoppix is the ultimate live CD. No geek-kit should be without it. Now Wim Vandersmissen has taken it a step futher by adding openMosix functionality. Drop the clusterKnoppix CD in your "server", boot up... boot up some networked clients... Knoppix built in LTSP magic kicks in and ta-da--instant cluster!"
I found out about OpenMosix recently, and I'd been looking for an excuse to test it out. This just makes it even easier.
I'm wondering how difficult it is to setup. Is it as easy as the poster made it sound?
Well, think of an environment where you have boxen sitting around unused part of the time, and want to be able to plug and unplug cluster components dynamically and not have any persistent data stored on the part time cluster members, Possibly even using them for windows and word processing during the day and cracking the xbox key at night.
Food not Bombs is a nice platitude but it breaks down when you notice that the Bombees are usually well fed
I wasnt trying to be funny I'm actually going to do this! It takes about 14 hours on a single windows machine right now (1Ghz AMDs). If I get 30 machines it should take little over half an hour to do a single movie. woo hoo
There is no god
"OpenMosixKnoppix didn't quite sound good, so I called it ClusterKnoppix ;)"
I would have chosen Kloppix...the "l" for cluster, the rest is self-explanitory.
What is the minimum hardware needed to support this? Obviously a NIC, but can it run diskless (no HDD or CD)?
:-)
That suddenly makes for a VERY cheap grid node. (Didn't want to use the "B" word
How about a non-copyright-infringing make-your-own-torrent web site? I'm sure it wouldn't get the bandwidth load that the other torrent sites. Then random people on slashdot could actually just do this themselves and post a comment with a torrent link.
Mad Software: Rantings on Developing So
"openMosix terminal server" - uses PXE, DHCP and tftp to boot linux clients via the network. No CDrom drive/harddisk/floppy needed for the clients
How do the clients work if no CDrom/HD/Floppy is needed? I am trying to wrap my brain around this one. I get the cluster server idea, but then does the server determine which clients on the network will boot into the cluster? Is it via DHCP? Doesn't there need to be *something* on the client side like a HD/floppy/CDrom so it can boot?
My beliefs do not require that you agree with them.
well I do actually.... GNU/Vidomi
There is no god
has replaced tomsrtbt as my rescue tool of choice.
..... what these people have managed to pull off is fantastic.
It probably would have done so even if any of my latest machines had a floppy drive
Je fume. Tu fumes. Nous fûmes!
You ain't from around these parts, are you? :-)
In all seriousness though, I do think that your MCSE and your Windows environment is limiting you here. I actually think the MCSE should be changed to CMSE, because you are a Certified Microsoft System Engineer. You are taught how to admin Microsoft systems only. It's OK, those are necessary things. But the problem is that you have been taught how to think in a "Microsoft world". There is a lot outside that world. Clustered computing is one of them. A bootable distro (ala Knoppix and others) is another.
I am sure when the bootable floppy distro came out, the MCSE's cried "what would I do with THAT?". Then CDRWs came about, and the bootable floppy turned into the bootable CD distro. The MS crowd said "Neat. Big deal." That has now turned into a bootable cluster server. Who knows where it might go from here. At some point, someone at Microsoft will say(or has already said) "Hey, that is cool. Can we do that?". They will try to buy the technology, and will find it can't be done. And they will try to build it from scratch, and there will be much wailing and gnashing of teeth.
I think it was Louis Armstrong, who when asked what Jazz is, said "Man, if you gotta ask, you'll never know." I am afraid that applies here.
My beliefs do not require that you agree with them.