Error calling R script from octave on HPC cluster

Question

I have one octave script which calls a R script to do some calculation on a HPC cluster. The calculation procedure is as follow:

1. Submit job on the cluster to get computation node assigned then distribute calculation to each CPUs in that node. Part of the shell script looks like this

   count=0
   
   HOSTLIST=
   
   for host in `cat $PBS_NODEFILE`
   
     do
   
       HOSTLIST[$count]=$host
   
       count=$(($count+1))
   
   done
   
   ...
   ...
   ...
   
   mkdir case_$count
   
   cd case_$count
   
   export workdir=`pwd`
   
   remotehost=${HOSTLIST[$pcount]}
   
   ssh -n $remotehost "cd $workdir; export PATH=$PATH:$workdir; octave $MFILE > /dev/null" &
   

2. For the sake of simplicity, the sample $MFILE content is

   printf("Calling R script from Octave \n");
   
   system('./hello_world.R');
   

3. The hello_world.R

   #!/usr/bin/Rscript
   print("Hello World!")
   

4. Error encountered when run

   sh: ./hellow_world.R: /usr/bin/Rscript: bad interpreter: No such file or directory
   

5. Some of my environment variables (just in case)

   $ echo $PATH
   
   /usr/local/bin:/bin:/usr/bin:/usr/local/sbin:/usr/sbin:/sbin:/usr/local/cuda/bin:/opt/ganglia/bin:/opt/ganglia/sbin:/usr/java/latest/bin:/opt/maven/bin:/opt/maui/bin:/opt/torque/bin:/opt/torque/sbin:/opt/pvfs2/bin:/opt/rocks/bin:/opt/rocks/sbin
   
   $ which Rscript 
   
   /usr/bin/Rscript
   
   $ which R
   
   /usr/bin/R
   

If I run the $MFILE from command line, it worked ok and printed the desired output just fine. I have tried many solutions I could find on the net to no avail

Anybody knows what went wrong? Thanks for any suggestion!

Answer 1

The problem was R was loaded on login node but wasn't loaded on the compute node. So in the job submission script, there must be a line to load R before doing any calculation. For example:

    module load r/3.4.3

See more here and here