diffusion.slurm
No OneTemporary
Actions

Subscribers

None

File Metadata

	#!/bin/bash
	# author: gilles foureste (EPFL)
	#
	#SBATCH --nodes=2
	# ntasks per node MUST be one, because multiple slaves per work doesn't
	# work well with slurm + spark in this script (they would need increasing
	# ports among other things)
	#SBATCH --ntasks-per-node=1
	#SBATCH --cpus-per-task=24
	#SBATCH --mem=8192
	# Beware! $HOME will not be expanded and invalid paths will result Slurm jobs
	# hanging indefinitely with status CG (completing) when calling scancel!
	##SBATCH --time=96:00:00
	#SBATCH --partition=scitas
	#SBATCH --qos=scitas
	#set -x
	#
	echo "---- starting $0 on $HOSTNAME"
	echo
	#
	MASTER_NODE=""
	start-spark.sh
	echo "configuration done..."
	set -x
	#
	#echo $MASTER_NODE
	#echo $MASTER
	#
	#MASTER_IP=$(cat ./sparklogs_${SLURM_JOBID}/spark_master)
	MASTER_IP=$(cat ${SLURM_JOBID}_spark_master)
	#
	echo $MASTER_IP
	time time spark-submit \
	--executor-memory 5G \
	--master $MASTER_IP \
	./diffusion.py
	#
	stop-spark.sh
	#scancel -u foureste
	#spark-submit --master $MASTER ./pi.py