installation help needed ...newbie
Many of your questions may have already been answered in earlier discussions or in the FAQ. The search results page will indicate current discussions as well as past list serves, articles, and papers.
Robert G. Brown rgb at phy.duke.eduSun Mar 10 05:48:59 PST 2002
- Previous message: installation help needed ...newbie
- Next message: Cheap pa-risc nodes
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
On Sat, 9 Mar 2002, rick bradshaw wrote: > I have a small setup with redhat7.2. I used lam and MPI and ssh. > here is a good link for using ssh without passwords > > http://www.cs.umd.edu/~arun/misc/ssh.html > > > and as for the other stuff like lam I setup the environment variable > > LAMRSH='ssh -x' > on all the machines. It seems to be working Just to add to this -- one major advantage of ssh is that it is AFAIK the only remote shell that allows an environment to be uniformly exported to target machines from the originating machines. If you put lines like PVM_ROOT=/usr/share/pvm3 PVM_RSH=/usr/bin/ssh XPVM_ROOT=/usr/share/pvm3/xpvm LAMRSH='ssh -x' in ~/.ssh/environment, it will add these variables to the environment on ALL target platforms when you ssh a command there. One might still wish for a global /etc/ssh/environment under root control, but root has other ways to accomplish the same thing I suppose. Believe it or not this can be incredibly useful. A lot of "user problems" with tools like pvm or lam occur when a user sets up the right environment on one host but the cluster extends onto nodes where the variables aren't correctly set and a user's normal home directory isn't mounted (and its .bashrc or .cshrc environment) are not executed. PVM has even tried to fix at least one recurring version of the problem by making the actual pvm command executed by the user a shell script that sets their PVM_ROOT correctly, but it doesn't set or forward PVM_RSH, so if you use ssh instead of the default rsh it often breaks anyway. One of rsh's biggest weaknesses is its lack of environment forwarding. I was struggling with crude hacks to "fix" this ten years ago (all of which were incredibly ugly and tended to break a lot). rgb -- Robert G. Brown http://www.phy.duke.edu/~rgb/ Duke University Dept. of Physics, Box 90305 Durham, N.C. 27708-0305 Phone: 1-919-660-2567 Fax: 919-660-2525 email:rgb at phy.duke.edu
- Previous message: installation help needed ...newbie
- Next message: Cheap pa-risc nodes
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Beowulf mailing list
