You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 34 Next »

SLURM

SLURM is new job scheduling system for the LCLS batch compute systems it is replacing the current LSF system. Generic documentation about SLURM can be found in this Quick Start User Guide. Even shorter documentation, some of it specific to psana, can be found in this page.

LSF to SLURM Cheat Sheets

Some quick guides showing equivalent commands in LSF and SLURM:

Partitions

The partition/queue information can be provided by the sinfo command.

sinfo

From the psana pool:

sinfo
psanagpu104:~$ sinfo
PARTITION  AVAIL  TIMELIMIT  NODES  STATE NODELIST
psanagpuq*    up 10-00:00:0      1 drain* psanagpu118
psanagpuq*    up 10-00:00:0      2  down* psanagpu[115-116]
psanagpuq*    up 10-00:00:0      1  drain psanagpu117
psanagpuq*    up 10-00:00:0      2   idle psanagpu[113-114]
psanaq        up 10-00:00:0      1 drain* psana1509
psanaq        up 10-00:00:0      4  down* psana[1503,1519,1604-1605]
psanaq        up 10-00:00:0      6    mix psana[1502,1504-1506,1520,1602]
psanaq        up 10-00:00:0      1  alloc psana1501
psanaq        up 10-00:00:0     27   idle psana[1507-1508,1510-1518,1601,1606-1620]
psanaq        up 10-00:00:0      1   down psana1603
The * following the name means default partition (queue) is psanagpuq

Check available GPUs on a specific node


scontrol show node psanagpu116 -d | grep Gres=gpu

From the psana pool:

Gres=gpu:1080ti:1(S:0)

Job Submission

There are 2 ways to submit a job on the cluster. The main way is by using the sbatch command for later execution, and the other is to submit an interactive job via srun. 

sbatch

The following is a simple submission script of a parallel psana batch job run with mpi.  It can be submitted with the command "sbatch submit.sh". The commands specified in the script file will be ran on the first available compute node that fits the resources requested.  There are two ideas: "nodes" and "tasks per node".  A "node" is a physical computer box (with a host-name, for example) but each box/node typically has multiple-cpu-cores (see this page for specific numbers:  Batch Nodes And Queues).  Typically the tasks-per-node parameter is set to utilize all the cores on each node.

> cat submit.sh
#!/bin/bash

#SBATCH --partition=psanaq
#SBATCH --nodes=2
#SBATCH --ntasks-per-node=3
#SBATCH --output=%j.log

# "-u" flushes print statements which can otherwise be hidden if mpi hangs
# "-m mpi4py.run" allows mpi to exit if one rank has an exception
mpirun python -u -m mpi4py.run /reg/g/psdm/tutorials/examplePython/mpiDataSource.py

One can also do this same command from the command line using the "--wrap" option for sbatch:

sbatch -p psanaq --nodes 2 --ntasks-per-node 3 --wrap="mpirun python mpi_simpletest.py"


This script shows some additional features controllable via SLURM:

sbatch
> cat tst_script 
#!/bin/bash
#
#SBATCH --job-name=<name> # Job name for allocation
#SBATCH --output=%j.log # File to which STDOUT will be written, %j inserts jobid
#SBATCH --error=%j.err # File to which STDERR will be written, %j inserts jobid
#SBATCH --partition=psanagpuq # Partition/Queue to submit job
#SBATCH --gres=gpu:1080ti:1 # Number of GPUs
#SBATCH --ntask=8  # Total number of tasks
#SBATCH --ntasks-per-node=4 # Number of tasks per node
#SBATCH --mail-user='username'@slac.stanford.edu # Receive e-mail from slurm
#SBATCH --mail-type=ALL # Type of e-mail from slurm; other options are: Error, Info.
#
srun -l hostname
srun python ExampleMultipleChaperones.py


> sbatch tst_script 
Submitted batch job 187
srun

Differently from sbatch, the srun command does not return immediately and waits for the job to complete. The srun command can be used to get control of a node to run interactively.  These can be useful for data exploration and software development.

The following are a few examples:

srun
>  srun -N2 -n4 hello.mpi 
Process 0 on psanagpu110 out of 1
Process 0 on psanagpu110 out of 1
Process 0 on psanagpu113 out of 1
Process 0 on psanagpu113 out of 1

Monitoring/Status

squeue

To check that jobs that exist on the system use the squeue command:

squeue
psanagpu104:~$ squeue
             JOBID PARTITION     NAME     USER ST       TIME  NODES NODELIST(REASON)
            466440    psanaq     wrap  khegazy  R 1-17:51:12      1 psana1502
            466423    psanaq     wrap  khegazy  R 1-17:53:31      1 psana1506
            466420    psanaq     wrap  khegazy  R 1-17:53:34      1 psana1602
            466421    psanaq     wrap  khegazy  R 1-17:53:34      1 psana1504

The ST (job state) field shows that jobid 466440 is currently running (R).

sacct

Get information about status of finished jobs

sacct
psanagpu104:~$ sacct -u khegazy
       JobID    JobName  Partition    Account  AllocCPUS      State ExitCode 
------------ ---------- ---------- ---------- ---------- ---------- -------- 
466416             wrap     psanaq                    10    RUNNING      0:0 
466416.batch      batch                               10    RUNNING      0:0 
466418             wrap     psanaq                    10    RUNNING      0:0 
466418.batch      batch                               10    RUNNING      0:0 
466420             wrap     psanaq                    10    RUNNING      0:0 
466420.batch      batch                               10    RUNNING      0:0



  • No labels