The bix people and IT group had a few discussions recently on how to use the cluster in LSB. This is a good attempt trying to find the strong business requirement to purchase new cluster instead of SMP sun big box.
This guide is excerpted from Chris Dwan (Bioteam). Original Link is here .
As my site has been linked by puzzlecat, MSN search engine regularly crawl this site, but M$ sucks and its search engine is just a crap. I decided to put my site into google by submitting the sitemap to it.
This is to follow up with parallel computing environment in R and I
implemented MPI for SNOW package as well. Here is the notes.
This notes shows how to install PVM, rpvm and SNOW to implement parallel computing for R environment.
when Ketan tried to load graph library, error occurs like
> library(graph)
Loading required package:
cluster
Error in assign(“nullgraphID”, getuuid()) :
couldn’t find function “getuuid”
Error: unable to load R code in package ‘graph’
Error: package/namespace load failed for ‘graph’
This is because Ruuid package could not be correctly loaded even there
is no warning. If we re-install Ruuid package, R will show the exit
code is not zero. The Ruuid installation failure is caused by
missing glib.h head file in Solaris. I have to install glib and
reinstall Ruuid package to fix it. Though the actual working version is
glib-1.2.9 (Ruuid use glib-config in PATH), the notes here shows installation for glib-2.6.5.
According LSB’s documents, this is the way to change SHELL.
On Solaris : passwd -r nis –e
On Linux : ypchsh wf002xx
(csh – /bin/csh ksh – /bin/ksh sh -
/bin/sh)
Important – Please enter only the path to the
configured shells or you will not able to login anymore after the change.
Though Aaron had great experience with unix and Mac, it is still difficult for him to understand the system in a few weeks. To help him better get the idea of the entire system, I spent a few days wrote a quick reference guide for TLL cluster administration.Hope it is useful for them.
The long waited Xserve Raid finally arrive with 8x400GB storage and it
will solve all our problems of being short of disk spaces. And the
installation is extremely easy.
Headnode image is Portal/wp_asr.dmg file. By default the headnode
platform is set to G5X, but we are using G4, so need to update it for
G4. To do it, edit Portal/rc.portal, add "echo G4 >
/Volumes/${VolmeName}/etc/bipod/platform" before reboot command.
Howerver the mod prevents iNQuiry from completing the installation in
the next boot. When you inputs information on each ethernet (though
appears right), it turns back to the initial welcome page, and lost in
the infinite loop.
Instead of using G4, I have to use the real G5 machines to test iNQuiry.