Difference between revisions of "Job submission"

From ScientificComputing
Jump to: navigation, search
Line 99: Line 99:
 
| Memory || -R "rusage[mem=2048]" || 1024 MB per core
 
| Memory || -R "rusage[mem=2048]" || 1024 MB per core
 
|-
 
|-
| Scratch space || -R "rusage[scratch=10000]"  
+
| Local scratch space || -R "rusage[scratch=10000]" || 0 MB
 
|}
 
|}
  

Revision as of 12:39, 12 August 2021

< Job management with LSF

Home

Submit a parallel job >


The three ways to access computing resources are

  1. Submit a job through command line
  2. Submit a job through a job bash script using #BSUB pragmas
  3. Get an interactive session on a compute node

Basic job submission

A basic BSUB job submission command consists of three parts:

bsub LSF options Job

where

bsub   is the LSF submit command.
LSF options   are for requesting resources and defining job-related options.
Job   is a computing job to be submitted.


Here is an example:

bsub -n 1 -W 4:00 -R "rusage[mem=4096]" "python myscript.py"

When the job is submitted, LSF shows job's information:

$ bsub -n 1 -W 4:00 -R "rusage[mem=4096]" "python myscript.py"
Generic job.
Job <8146539> is submitted to queue <normal.4h>

The output includes

  1. Job type, e.g., Generic Job, MPI Job or Abaqus Job
  2. Unique JobID, e.g., 8146539
  3. The queue, e.g., normal.4h, normal.24h, or normal.120h

Note: The JobID is important for monitoring or reporting issues. It is also integrated in the name of the LSF output file. Please don't delete the LSF output file unless you are sure that the job was running fine.

Job

A job can be one of the following:

Job Command Examples of job submission command
a single Linux command cmd
a program with its path /path/to/myprogram bsub ./bin/hello
a command or program with its arguments cmd arg1 arg2 bsub echo hello
multiple commands "cmd1 ; cmd2" bsub "date; pwd; ls -l"
piped command "cmd1 | cmd2"
a command with I/O redirection, quote "cmd<in >out" bsub "du -sk /scratch > du.out"
a here document, passed via "<<" << EOF ... EOF
a shell script, passed via "<" < script bsub < hello.sh


LSF options

Requesting resources

Resources Format Default values
Maximum run time -W HH:MM 04:00 (4 hours)
Number of processors -n nprocs 1 processor
Memory -R "rusage[mem=2048]" 1024 MB per core
Local scratch space -R "rusage[scratch=10000]" 0 MB

Other LSF options

-o outfile append job’s standard output to outfile
-e errfile append job’s error messages to errfile
-R "rusage[...]" advanced resource requirement (memory,...)
-J jobname assign a jobname to the job
-w "depcond" wait until dependency condition is satisfied
-Is submit an interactive job with pseudo-terminal
-B /-N send an email when the job begins/ends
-u user@domain use this address instead of username@ethz.ch

LSF submission line advisor can assist your to find LSF options you need.

Job script and #BSUB pragmas

Create a job script called job_script.bsub

#!/bin/bash
#BSUB -n 24                     # 24 cores
#BSUB -W 8:00                   # 8-hour run-time
#BSUB -R "rusage[mem=4000]"     # 4000 MB per core
#BSUB -J analysis1
#BSUB -o analysis1.out
#BSUB -e analysis1.err
#BSUB -N

module load gcc/6.3.0 openmpi/3.0.2
cd /path/to/execution/folder
mpirun myprogram arg1

Submit a job

$ bsub < job_script.bsub

Interactive session on a compute node

To run a quick test or a benchmark, you can request an interactive session on a compute node by using the BSUB option

-I, -Ip or -Is

For example:

[jarunanp@eu-login-38 ~]$ bsub -n 4 -W 01:00 -Is bash
Generic job.
Job <161197292> is submitted to queue <normal.4h>.
<<Waiting for dispatch ...>>
<<Starting on eu-ms-001-15>>
[jarunanp@eu-ms-001-15 ~]$

Further reading

Helper


< Job management with LSF

Home

Submit a parallel job >