job_pipeline.sh
fda646ff333amaster

job_pipeline.sh
View Options

			#! /bin/bash
			#This script automatizes the launch of multiple job with chain dependency.
			nu_=0.01
			dnu=0.01

			Tm_=2000
			dTm=2000

			# First submit
			lastji=$(sbatch submit_00.cmd)
			lastjid=${lastji##* }
			echo $lastji
			#echo $lastjid

			for i in {1..1}; do
			# Setup indices of job id (current and previous one)
			im1=$(awk "BEGIN {print $i-1}")
			idm1=$(printf '%02d' $im1)
			id=$(printf '%02d' $i)

			# Create new submit file from older one
			awk -v "ID=$id" '{
			if (NR == 8) print "#SBATCH --error=err_"ID".txt";
			else if (NR == 9) print "#SBATCH --output=out_"ID".txt";
			else if (NR == 12) print "srun --cpu-bind=cores ./gyacomo 2 24 1 "ID;
			else print $0}' submit_$idm1.cmd > submit_$id.cmd

			# Create new fort file from older one
			awk -v "NU=$nu_" -v "TM=$Tm_" -v "J2L=$im1" '{
			if (NR == 04) print " tmax = "TM;
			else if (NR == 40) print " job2load = "J2L;
			else if (NR == 54) print " nu = "NU;
			else print $0}' fort_$idm1.90 > fort_$id.90

			# Retrieve last jobid and launch next job with dep
			lastji=$(sbatch --dependency=afterok:$lastjid submit_0$i.cmd)
			lastjid=${lastji##* }
			echo $lastjid

			# Increment variables
			nu_=$(awk "BEGIN {print $nu_+$dnu}")
			Tm_=$(awk "BEGIN {print $Tm_+$dTm}")
			done