OpenMosix Conference Delves Into Clustering
axehind writes "There's a article on Newsforge about the conference titled "Linux Cluster: the openMosix Approach" that took place in Italy on Nov 28th. It's really interesting to see what openMosix clusters have been used for. From game clusters to scientific research . It includes links to the conferences slides and some of the papers."
A cabinet of coyote not same as a bay of wolf
Can you imagine a Beowulf cluster of these?
Error: PANTS NOT FOUND. Press <F1> to continue.
The link to the slides includes a nifty detection script that informs me that I don't have Windows Media Player installed as one of the plugins for Mozilla.
Where do I get a free implementation of WMP for Linux?
"Provided by the management for your protection."
Step 1: Link to goatse
Step 2.-1: In Soviet Russia all your first post are belong to beowulf cluster
Step 2: ???
Step 2.1: Point out that this story's already been posted
Step 2.2: Don't read article
Step 2.3: Become speling/grammer Nazi
Step 2.4: Focus on minor inconsistencies and ignore main point
Step 3: Profit
Step 4: (+5, Troll) I win!
Step 4.1: (-1, Informative) I win again!
Step 5: Submit to Slashdot as "How to Troll on Slashdot"
One of the slides talked about migrating processes with threads and their plans to have distributed memory support in by summer 2003, but to my surprise there was no talk about migrating processes with open sockets (without dropping the connections of course). Is this a low priority for OpenMOSIX?
on a 1.3Ghz and an 800Mhz Athlon. Very neat stuff. I needed to recompile ALSA, so I used that as a test. With clustering disabled the build time was 2min and 51 sec. With clustering enabled it took 1min and 51.
The only thing I didn't like was having to rollback to a 2.4.19 kernel, because I've been using 2.4.20 and its stable, but I guess there'll be a new version eventually.
Vote for global prefs bug
... unless the howto is out of date, distributed shared memory support still needs to be added. I would think that this would be the top priority, particularly since any DB clusters would have to have this.
Hey IBM, how about it?