On Wed, Nov 05, 2008 at 06:48:42PM -0500, Rodrick Brown wrote:
> X-Spam-Level: 
> X-Spam-Status: No, score=0.001 tagged_above=-5 required=6.31
>       tests=[HTML_MESSAGE=0.001]
> From: Rodrick Brown <[EMAIL PROTECTED]>
> To: SAGE mailing list <[EMAIL PROTECTED]>,
>       "<[email protected]>" <[email protected]>
> Subject: [lopsa-tech] Parallel Execution of Scripts and Applications
> 
> I'm looking for something that can execute jobs across a large number of
> hosts mostly adhoc running of various processes. Scheduling would be nice
> too but not necessary right now.
> I previously used a tool called gexecd part of the ganglia suite of
> applications but I've found numerous limitations with this tool when running
> applications through this process.
> Currently I've reverted back to running my jobs through ssh but this doesn't
> seem like the best approach and was wondering what everyone else is doing
> for starting and stopping applications across a large network of servers.

In addition to everything else also mentioned, there's pdsh:
https://computing.llnl.gov/linux/pdsh.html

basically a port of IBM's dsh to linux, iirc.

Has some neat features, e.g.:

pdsh -w "node[1-4]" ls -la /etc/services

Will give nicely formatted output as such:

----------------
node[1-4]
----------------
 -rw-r--r-- 1 root root 362031 Feb 23  2006 /etc/services

grouping the like output from different nodes.

pdsh has multiple methods you can choose from, including ssh, rsh, anda
few others.

You could also consider using SLURM (also from Livermore) to distribute
those jobs.  I know of a vendor that does something similar.  It might
be more overhead than you want.

-jkl
_______________________________________________
Tech mailing list
[email protected]
http://lopsa.org/cgi-bin/mailman/listinfo/tech
This list provided by the League of Professional System Administrators
 http://lopsa.org/

Reply via email to