[gridengine users] Kicking off servers via Grid Engine

Earl Lazarus earl.lazarus at gmail.com
Tue Apr 3 17:14:12 UTC 2012

Is "prologue script" a Grid Engine concept?  What I am doing right now is
running a script that submits jobs to SGE to set up a list of environment
servers on a list of hosts.  Each job sets up one environment on one
host...these jobs occupy a slot unfortunately and limit the number of
simulations I can run on each host.  In fact if the number of environments
running equals the number of slots/CPUs, then I cannot run ANY
simulations.  How does a "prologe script" get around this?

On Tue, Apr 3, 2012 at 12:03 PM, Jesse Becker <beckerjes at mail.nih.gov>wrote:

> How about using a prologue script to setup your environment before
> running the job?
> On Tue, Apr 03, 2012 at 12:56:47PM -0400, Earl Lazarus wrote:
>> Our environment is a linux cluster with 4 CPUs per machine.  We are
>> changing the architecture of a Monte Carlo simulation that we routinely run
>> via Grid Engine.  In the new paradigm, each instance will need to
>> communicate via sockets with an "Environment Server" running on the host
>> where the sim executes.  We need to kick off these environment servers
>> prior to submitting the simulation jobs.  These servers are basically
>> inactive unless queried by one of the sims.  We also want to allow the
>> environment servers to run possibly for days or weeks at a time as they are
>> low impact.
>> However, what I really want to do is kick off an environment server on a
>> host and not have it count against the "slot count" for that host.  There
>> will be instances where I might need 4 different environment servers
>> running on each host.  Unfortunately that would occupy all of my slots and
>> NONE of my simulations would ever run!!
>> I have seen the concept of "Parallel Environments" in the SGE User
>> Manual, but the explanation is pretty much worthless.
>> Bottom line...I need to pre-position some executables on each host
>> without consuming slots.   Note that I tried to kick them off in a script
>> of the form:
>> sppe &
>> exit
>> but Grid Engine monitors jobs placed in the background and kills them!
>> The possibility of avoiding Grid Engine entirely and kicking them off
>> with rsh/ssh is something we want to avoid.
>> earl
>  ______________________________**_________________
>> users mailing list
>> users at gridengine.org
>> https://gridengine.org/**mailman/listinfo/users<https://gridengine.org/mailman/listinfo/users>
> --
> Jesse Becker
> NHGRI Linux support (Digicon Contractor)
> :(){ :&:};:
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://gridengine.org/pipermail/users/attachments/20120403/06a041eb/attachment.html>

More information about the users mailing list