When I submit a job to Condor-C using a pre-staged executable and the following submit file: ------------------------------------------------ Universe = grid
Executable = /usr/local/bin/casal Arguments = -e -O mpd.dat
Error = Error.txt Output = Output.txt Log = Condor.txt
transferinputfiles = estimation.csl, population.csl, output.csl, mpd.dat shouldtransferfiles = True whentotransferoutput = ON_EXIT transfer_executable = False
grid_resource = condor erm-43880.yyy.zzz erm-43880.yyy.zzz +remote_requirements = Arch == "X86_64" && OpSys == "LINUX" +remote_universe = vanilla +remote_shouldtransferfiles = "YES" +remote_whentotransferoutput = "ON_EXIT"
Queue -----------------------------------------------
The execute machine is failing because it is trying to find the executable in the spool directory:
6/6 03:48:48 About to exec /opt/condor-6.9.3/local.ERM-43880/spool/cluster7.proc0.subproc0/casal -e -O mpd.dat 6/6 03:48:48 Env = _CONDOR_SCRATCH_DIR=/opt/condor-6.9.3/local.amlr-svr4/execute/dir_8572 _CONDOR_SLOT=1 6/6 03:48:48 Create_Process: Cannot access specified executable "/opt/condor-6.9.3/local.ERM-43880/spool/cluster7.proc0.subproc0/casal": errno = 2 (No such file or directory) 6/6 03:48:48 ERROR "Create_Process(/opt/condor-6.9.3/local.ERM-43880/spool/cluster7.proc0.subproc0/casal,-e -O mpd.dat, ...) failed" at line 420 in file os_proc.C
Is this a bug? Or is there a way around the problem, without doing a file transfer on the executable every time? ___________________________________________________________________________ Australian Antarctic Division - Commonwealth of
Australia |