Hello, right now I'm thinking about running R 2.0.0 on box A (Debian SID) but at the same time having access to the ressources of box B (Ubuntu Linux) regarding disk capacity, RAM, idle CPU cycles . Is there anyone of you that has already installed and administred such a tiny (home based) cluster? Well, the term cluster seems quite a bit of an exaggeration related to what I am looking for! I am absolutely new to this subject and I would appreciate some hints on where to start doing such things with my GNU R on both boxes. Maybe a pointer too some introductory materials focussed to doing clustering stuff with GNU R would be sufficient. Are there any recommended methods or tools to realize a two nodes cluster? regards Thomas
Using GNU R on a two box "cluster"
4 messages · Dan Bolser, Dirk Eddelbuettel, Thomas Friedrichsmeier
On Sun, 7 Nov 2004, [ISO-8859-15] Thomas Schönhoff wrote:
Hello, right now I'm thinking about running R 2.0.0 on box A (Debian SID) but at the same time having access to the ressources of box B (Ubuntu Linux) regarding disk capacity, RAM, idle CPU cycles . Is there anyone of you that has already installed and administred such a tiny (home based) cluster? Well, the term cluster seems quite a bit of an exaggeration related to what I am looking for! I am absolutely new to this subject and I would appreciate some hints on where to start doing such things with my GNU R on both boxes. Maybe a pointer too some introductory materials focussed to doing clustering stuff with GNU R would be sufficient. Are there any recommended methods or tools to realize a two nodes cluster?
The absolute *best* source of cluster information anywhere on the web (subjectively speaking) is the bioclusters mailing list... http://bioinformatics.org/lists/bioclusters The list has a surfeit of cluster experts who make a living from consultancy.
regards Thomas
______________________________________________ R-help at stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
On Sun, Nov 07, 2004 at 11:42:33AM +0100, Thomas Sch?nhoff wrote:
right now I'm thinking about running R 2.0.0 on box A (Debian SID) but at the same time having access to the ressources of box B (Ubuntu Linux) regarding disk capacity, RAM, idle CPU cycles . Is there anyone of you that has already installed and administred such a tiny (home based) cluster?
[...]
Are there any recommended methods or tools to realize a two nodes cluster?
The best place to start is probably the high-level snow package which does all clustering in userspace -- i.e. it'll work on your machines just by pulling in the packages it needs. Moreover, as both of your machines are Debian-based, all you need should $ apt-get install r-cran-snow modulo maybe deciding which communications protocol you want to use (pvm or lam/mpi). Another approach is to burn a dvd from the Quantian iso images (see http://dirk.eddelbuettel.com/quantian/) and boot one machine with it (note that there are also some tricks that help you do that without burning a dvd, but that is still the best way to start). You can then boot a second machine rather easily by starting the openMosix Terminalserver that is part of Quantian --- and you second machine can be booted off the first one already running Quantian (provided is can boot via PXE, many recent BIOS support this). That gives you a two-node cluster with openMosix, and you can start, say, two R sessions with lengthy simulations, and one should migrate automagically to other box. Hth, Dirk
If your hair is standing up, then you are in extreme danger.
-- http://www.usafa.af.mil/dfp/cockpit-phys/fp1ex3.htm
Hello, Dirk Eddelbuettel schrieb:
On Sun, Nov 07, 2004 at 11:42:33AM +0100, Thomas Sch?nhoff wrote:
right now I'm thinking about running R 2.0.0 on box A (Debian SID) but at the same time having access to the ressources of box B (Ubuntu Linux) regarding disk capacity, RAM, idle CPU cycles . Is there anyone of you that has already installed and administred such a tiny (home based) cluster?
[...]
Are there any recommended methods or tools to realize a two nodes cluster?
The best place to start is probably the high-level snow package which does all clustering in userspace -- i.e. it'll work on your machines just by pulling in the packages it needs. Moreover, as both of your machines are Debian-based, all you need should $ apt-get install r-cran-snow modulo maybe deciding which communications protocol you want to use (pvm or lam/mpi). Another approach is to burn a dvd from the Quantian iso images (see http://dirk.eddelbuettel.com/quantian/) and boot one machine with it (note that there are also some tricks that help you do that without burning a dvd, but that is still the best way to start). You can then boot a second machine rather easily by starting the openMosix Terminalserver that is part of Quantian --- and you second machine can be booted off the first one already running Quantian (provided is can boot via PXE, many recent BIOS support this). That gives you a two-node cluster with openMosix, and you can start, say, two R sessions with lengthy simulations, and one should migrate automagically to other box.
Thanks to you and Dan for response, seems like a lot to be done ;-) sincerely Thomas