I did:
********************
[root@malambo src]# cexec switcher mpi = mpich-1.2.4 --system
************************* oscar_cluster *************************
processing node node2.ece.uprm.edu
processing node node3.ece.uprm.edu
processing node node4.ece.uprm.edu
processing node node5.ece.uprm.edu
processing node node6.ece.uprm.edu
processing node node7.ece.uprm.edu
processing node node8.ece.uprm.edu
******
I went to eat, later I came back and the computer still like up,
for that reason I break ^c, and it showed me the following:
Keyboard interrupt
Keyboard interrupt
Keyboard interrupt
Keyboard interrupt
Keyboard interrupt
Keyboard interrupt
Keyboard interrupt
Keyboard interrupt
**********
What hapenned? :(
The cluster has 8 nodes, one server and 7 clients
and I repeat the following:
*****************
[root@malambo src]# ssh node3
Last login: Fri Nov 15 01:26:46 2002 from malambo.ece.uprm.edu
switcher:mpi: Cannot find modulefile for mpic-1.2.4 -- skipping
Daneil Burbano
Puerto Rico University at Mayaguez
**********************************************************
On Fri, 15 Nov 2002, Jeff Squyres wrote:
> On Fri, 15 Nov 2002, Daniel Alberto Burbano wrote:
>
> > As root, where the server is malambo:
> >
> > ****************
> > [root@malambo root]# ssh malambo
> > Warning: Permanently added 'malambo,192.168.1.200' (RSA) to the list of
> > known hosts.
> > Last login: Fri Nov 15 00:31:34 2002
> > ****************
>
> This is ok -- this always happens the first time you ssh somewhere.
>
> > as root, where the first client is node2:
> >
0> > ************
> > [dburbano@malambo dburbano]$ ssh node2
> > Last login: Fri Nov 15 01:38:46 2002 from malambo.ece.uprm.edu
> > switcher:mpi: Cannot find modulefile for mpic-1.2.4 -- skipping
> > [dburbano@node2 dburbano]$
> > ****************
>
> As you discuss below, the problem is because you don't have an MPI
> implementation by that name. Hence, switcher complains.
>
> > I have the message of switcher, because I don't have an MPI
> > implementation with that name, I canged it to available MPI
> > implementation and I did a manual execute to push the configuration
> > information to the nodes, but it did not upgrate. The command was:
> >
> > /opt/opium/bin/sync_users
>
> Yes, this is an unfortunate catch-22. If you set an incorrect MPI and
> then push it out, you're hosed. I see two ways to fix this:
>
> 1. Don't push an incorrectly-set MPI out. ;-)
> 2. Use "cexec" instead of "cpush" to set the right MPI. cexec will run a
> command on every node instead of pushing files out, so it won't run into
> the same rsync issues:
>
> cexec switcher mpi = mpich-1.2.4 --system
>
> That should set it correctly on all machines (such that sync_users isn't
> necessary), and then cpush will work properly in the future.
>
> --
> {+} Jeff Squyres
> {+} [EMAIL PROTECTED]
> {+} http://www.lam-mpi.org/
>
>
>
> -------------------------------------------------------
> This sf.net email is sponsored by: To learn the basics of securing
> your web site with SSL, click here to get a FREE TRIAL of a Thawte
> Server Certificate: http://www.gothawte.com/rd524.html
> _______________________________________________
> Oscar-users mailing list
> [EMAIL PROTECTED]
> https://lists.sourceforge.net/lists/listinfo/oscar-users
>
-------------------------------------------------------
This sf.net email is sponsored by: To learn the basics of securing
your web site with SSL, click here to get a FREE TRIAL of a Thawte
Server Certificate: http://www.gothawte.com/rd524.html
_______________________________________________
Oscar-users mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/oscar-users
