Re: [Condor-users] running namd in Condor-G

["emir mahmut bahsi" <embahsi@xxxxxxxxx>]

I still could not solve the following problem. However, I have founded exactly what the problem is. Condor does not run the job with the argument that I specify. In order to get the same error that condor produces I tried the following command line(intentionally):

mpirun -np $np -machinefile $PBS_NODEFILE namd2  #min1.conf (I am not providing the configuration file in order to get the error).

I am getting:

FATAL ERROR: No simulation config file specified on command line.

It is the same error that I am getting with condor. That means condor is submitting the job without giving configuration file as a command line argument. However, I have already mentioned the configuration file in the submit file as arguments=... (you can see the configuration file in the forwarded email).

Does anybody has any idea about that?

Thanks,
Emir

On Thu, May 15, 2008 at 4:50 PM, emir mahmut bahsi <embahsi@xxxxxxxxx> wrote:

Hello all,
I am trying to submit a condor job in my local machine which submits it to a site that has pbs scheduler. This job is an mpirun job and it runs namd2 via mpi.
namd2 is one executable which is used in some (in biology) scientific calculations. namd2 takes a file path as an argument.

Command line e.g. : namd2 min1.conf
Command line e.g for mpi-namd : mpirun -np $np -machinefile $PBS_NODEFILE namd2  min1.conf

However, whatever I try it does not except that argument as the file name. It always gives me the following error:

------------- Processor 0 Exiting: Called CmiAbort ------------
Reason: FATAL ERROR: No simulation config file specified on command line.

My condor submit file is following:

universe     = globus
globusscheduler = <site>/jobmanager-pbs
executable   = <directory path for namd>/namd2
arguments  = "<directory path for config file>/min1.conf"
environment = "CHARMRUN=<directory path for charmrun>/charmrun"
remote_initialdir = /.../../.../sim/0000
should_transfer_files = YES
when_to_transfer_output = ON_EXIT
transfer_executable = false
error        = tom_condor_old.error
output       = tom_condor_old.output
log          = tom_condor_old.log
notification = never
globusrsl = (project=<allocation>)(jobType=mpi)(count=04)(queue = checkpt)(arguments=<directory path for config file>/min1.conf)
queue

Since I did not know whether I have to give the argument name as the argument in condor-dagman or argument as a value in globusrsl; I wrote in both places (as you can see above).

I am trying to solve that problem for days but I could not.

Thanks,
--
Emir Mahmut Bahsi
Research assistant
Dept of Computer Science at LSU
Room 231, Johnston Hall
Louisiana State University
Baton Rouge, LA 70803
embahsi@xxxxxxxxxxx
http://www.cct.lsu.edu/~embahsi