Hi Ross,
Indeed I tried to use the script from a pbs shell and it only worked with
the slight modifications I've put in there.
I tried to run the script on my local SGI machine (from the prompter)
and it runs fine.
You are indeed right, it has to do with the fact that the
processes are shifted into the background.
My point is that if you want to "&" your process you also need to use
crontab or set up a way in the shell script to check regularly for the
processes that a certain user has running in the queue.
I think crontab is not such a bad idea if you only count the processes
submitted by a certain user and also use a specific keyword for the
description of the job that you want to follow later on.
Best,
Ioana
On Thu, 8 Apr 2004, Ross Walker wrote:
> Dear Ioana,
>
> I wrote this script and on the machine I tested it on it runs fine (P4 3GHz
> Redhat 9.0). I just tested it again on one of our cluster machines and it
> also runs correctly, both on one processor or on 4 processors (using a
> slightly modified script that calls mpirun). This is whether I run it
> interactively or in the background (./run_a-dna_20-1820ps.x &).
>
> Do you have sander set up in some strange way that means it immediately
> backgrounds itself upon execution? If you run sander manually on the command
> line does it return the command prompt straight away or after the
> calculation has finished?
>
> If you could send me the details of the exact command you are running, a
> copy of the script you use (in case it is in some way different to the one
> that is on the website) and a list of any screen output / error messages you
> receive that would make it much easier for me to determine what is going on.
>
> Note, if you are submitting this script to a batch queuing system such as
> PBS there is no guarantee it will run correctly. This will depend on how
> your queue system is set up. In this case it is probably best to speak to
> the people who manage the queue so that they can advise you on the form of
> run script to use.
>
> Ps. Using crontab is a very bad idea since you have know way of knowing how
> long the previous job will take. E.g if the machine gets loaded with other
> jobs while yours is running you might end up starting new jobs before the
> previous one finished. This would undoubtedly corrupt your results.
>
> All the best
> Ross
>
> /\
> \/
> |\oss Walker
>
> | Department of Molecular Biology TPC15 |
> | The Scripps Research Institute |
> | Tel:- +1 858 784 8889 |
> | EMail:- ross.rosswalker.co.uk |
> | http://www.rosswalker.co.uk/ | PGP Key available on request |
>
>
>
> > -----Original Message-----
> > From: owner-amber.scripps.edu
> > [mailto:owner-amber.scripps.edu] On Behalf Of Ioana Cozmuta
> > Sent: 08 April 2004 19:26
> > To: Amber List
> > Subject: AMBER: script on tutorial page
> >
> > Hi amber users,
> >
> > I think there is a mistake in the script
> > run_a-dna_20-1820ps.x
> > on the tutorial web site:
> >
> > http://amber.scripps.edu/tutorial/polyA-polyT_New/adna.html
> >
> > The text says that the total 1.8ns run will be executed by
> > running 9 x 200
> > ps simulations with each successive
> > simulation continuing on from the previous one.
> > The script submits all the 9 simulations at the same time and not
> > successively!
> >
> > Ioana
> >
> > Here is what I would propose:
> >
> > use crontab to run the script.csh every 30 minutes:
> > 30 * * * * $PATH/script.csh
> >
> > Content of:
> > script.csh
> >
> >
> > ****************************************
> > #!/bin/csh
> > set AMBERHOME="/usr/local/AMBER8"
> > set MDSTARTJOB=2
> > set MDENDJOB=10
> > set MDCURRENTJOB=$MDSTARTJOB
> > set MDINPUT=0
> > echo -n "Starting Script at: "
> > date
> > echo ""
> >
> > set user = `whoami`
> >
> > while ( $MDCURRENTJOB <= $MDENDJOB )
> > set chk = `ps -u $user |grep sander | wc -l`
> > if ($chk == 0) then
> > echo -n "Job $MDCURRENTJOB started at: "
> > date
> > . MDINPUT = $MDCURRENTJOB - 1
> > $AMBERHOME/exe/sander -O -i a-dna_md_1800ps.in \
> > -o
> > a-dna_md$MDCURRENTJOB.out \
> > -p a-dna_wat.prmtop \
> > -c a-dna_md$MDINPUT.rst \
> > -r
> > a-dna_md$MDCURRENTJOB.rst \
> > -x
> > a-dna_md$MDCURRENTJOB.mdcrd
> > gzip -9 -v a-dna_md$MDCURRENTJOB.mdcrd
> > echo -n "Job $MDCURRENTJOB finished at: "
> > date
> > . MDCURRENTJOB = $MDCURRENTJOB + 1
> > endif
> > end
> > echo "ALL DONE"
> >
> >
> > ******************
> > Ioana Cozmuta, PhD
> > Research Scientist
> > Eloret Corporation
> > Mail Stop 230-3
> > NASA AMES Research Center,
> > Moffet Field, CA 94035-1000
> > Phone: (650) 604-0993
> > http://www.ipt.arc.nasa.gov/computationalpeople.html
> > --------------------------------------------------------------
> > ---------
> > The AMBER Mail Reflector
> > To post, send mail to amber.scripps.edu
> > To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
> >
>
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber.scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
>
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Fri Apr 09 2004 - 23:53:00 PDT
