Slurm提交MPI作业
Slurm提交MPI作业
首先准备一个MPI程序，这里使用python语言的mpi4py库写了一个
helloworld.py
#!/usr/bin/env python
"""
Parallel Hello World
"""
from mpi4py import MPI
import sys
import time
size = MPI.COMM_WORLD.Get_size()
rank = MPI.COMM_WORLD.Get_rank()
name = MPI.Get_processor_name()
sys.stdout.write("Hello, World! I am process %d of %d on %s.\n" % (rank, size, name))
time.sleep(300)
Slurm提交作业脚本
helloworld.sh
#!/bin/sh
#SBATCH -o /apps/mpi/myjob.out
#SBATCH --nodes=2
#SBATCH --ntasks-per-node=2
mpirun python /apps/mpi/helloworld.py
Slurm提交MPI作业
$ sbatch helloworld.sh查看MPI作业信息
查看MPI作业状态
$ squeue
JOBID PARTITION NAME USER ST TIME NODES NODELIST(REASON)
40 control hellowor jhadmin R 3:06 2 centos6x[1-2]查看MPI作业详细信息
$ scontrol show jobs
JobId=40 JobName=helloworld.sh
UserId=jhadmin(500) GroupId=jhadmin(500) MCS_label=N/A
Priority=4294901724 Nice=0 Account=(null) QOS=(null)
JobState=COMPLETED Reason=None Dependency=(null)
Requeue=1 Restarts=0 BatchFlag=1 Reboot=0 ExitCode=0:0
RunTime=00:05:01 TimeLimit=UNLIMITED TimeMin=N/A
SubmitTime=2016-09-12T04:27:00 EligibleTime=2016-09-12T04:27:00
StartTime=2016-09-12T04:27:00 EndTime=2016-09-12T04:32:01 Deadline=N/A
PreemptTime=None SuspendTime=None SecsPreSuspend=0
Partition=control AllocNode:Sid=centos6x1:2239
ReqNodeList=(null) ExcNodeList=(null)
NodeList=centos6x[1-2]
BatchHost=centos6x1
NumNodes=2 NumCPUs=4 NumTasks=4 CPUs/Task=1 ReqB:S:C:T=0:0:*:*
TRES=cpu=4,node=2
Socks/Node=* NtasksPerN:B:S:C=2:0:*:* CoreSpec=*
MinCPUsNode=2 MinMemoryNode=0 MinTmpDiskNode=0
Features=(null) Gres=(null) Reservation=(null)
OverSubscribe=OK Contiguous=0 Licenses=(null) Network=(null)
Command=/apps/mpi/helloworld.sh
WorkDir=/apps/mpi
StdErr=/apps/mpi/myjob.out
StdIn=/dev/null
StdOut=/apps/mpi/myjob.out
Power=MPI输出信息
$ cat /apps/mpi/myjob.out
srun: cluster configuration lacks support for cpu binding
Hello, World! I am process 0 of 4 on centos6x1.
Hello, World! I am process 1 of 4 on centos6x1.
Hello, World! I am process 2 of 4 on centos6x2.
Hello, World! I am process 3 of 4 on centos6x2.作业进程信息
centos6x1
pstree -apl 6290
slurmstepd,6290
├─slurm_script,6294 /tmp/slurmd/job00040/slurm_script
│ └─mpirun,6295 python /apps/mpi/helloworld.py
│ ├─python,6306 /apps/mpi/helloworld.py
│ │ └─{python},6309
│ ├─python,6307 /apps/mpi/helloworld.py
│ │ └─{python},6308
│ ├─srun,6297 --ntasks-per-node=1 --kill-on-bad-exit --cpu_bind=none --nodes=1 --nodelist=centos6x2 --ntasks=1 orted -mca orte_ess_jobid37944
│ │ ├─srun,6300 --ntasks-per-node=1 --kill-on-bad-exit --cpu_bind=none --nodes=1 --nodelist=centos6x2 --ntasks=1 orted -mca orte_ess_jobid37944
│ │ ├─{srun},6301
│ │ ├─{srun},6302
│ │ └─{srun},6303
│ └─{mpirun},6296
├─{slurmstepd},6292
└─{slurmstepd},6293centos6x2
pstree -apl 4655
slurmstepd,4655
├─orted,4660 -mca orte_ess_jobid 3794403328 -mca orte_ess_vpid 1 -mca orte_ess_num_procs 2 -mca orte_hnp_uri"3794403
│ ├─python,4663 /apps/mpi/helloworld.py
│ │ └─{python},4665
│ └─python,4664 /apps/mpi/helloworld.py
│ └─{python},4666
├─{slurmstepd},4657
├─{slurmstepd},4658
└─{slurmstepd},4659另一种方式提交MPI作业
$ salloc -n 8 mpiexec python /apps/mpi/helloworld.py
...
Hello, World! I am process 1 of 8 on centos6x1.
Hello, World! I am process 0 of 8 on centos6x1.
Hello, World! I am process 3 of 8 on centos6x1.
Hello, World! I am process 2 of 8 on centos6x1.
Hello, World! I am process 4 of 8 on centos6x2.
Hello, World! I am process 6 of 8 on centos6x2.
Hello, World! I am process 7 of 8 on centos6x2.
Hello, World! I am process 5 of 8 on centos6x2.作业进程信息
centos6x1
$ pstree -apl 8212
salloc,8212 -n 8 mpiexec python /apps/mpi/helloworld.py
├─mpiexec,8216 python /apps/mpi/helloworld.py
│ ├─python,8227 /apps/mpi/helloworld.py
│ │ └─{python},8231
│ ├─python,8228 /apps/mpi/helloworld.py
│ │ └─{python},8232
│ ├─python,8229 /apps/mpi/helloworld.py
│ │ └─{python},8233
│ ├─python,8230 /apps/mpi/helloworld.py
│ │ └─{python},8234
│ ├─srun,8218 --ntasks-per-node=1 --kill-on-bad-exit --cpu_bind=none --nodes=1 --nodelist=centos6x2 --ntasks=1 orted -mca orte_ess_jobid36682
│ │ ├─srun,8221 --ntasks-per-node=1 --kill-on-bad-exit --cpu_bind=none --nodes=1 --nodelist=centos6x2 --ntasks=1 orted -mca orte_ess_jobid36682
│ │ ├─{srun},8222
│ │ ├─{srun},8223
│ │ └─{srun},8224
│ └─{mpiexec},8217
└─{salloc},8213centos6x2
$ pstree -apl 6356
slurmstepd,6356
├─orted,6369 -mca orte_ess_jobid 3668246528 -mca orte_ess_vpid 1 -mca orte_ess_num_procs 2 -mca orte_hnp_uri"3668246
│ ├─python,6372 /apps/mpi/helloworld.py
│ │ └─{python},6376
│ ├─python,6373 /apps/mpi/helloworld.py
│ │ └─{python},6378
│ ├─python,6374 /apps/mpi/helloworld.py
│ │ └─{python},6377
│ └─python,6375 /apps/mpi/helloworld.py
│ └─{python},6379
├─{slurmstepd},6366
├─{slurmstepd},6367
└─{slurmstepd},6368转载请以链接形式标明本文链接
本文链接：http://blog.csdn.net/kongxx/article/details/52592677