Slurm - Revision history

Test admin: Move content for slurm to an obsolete page

2022-08-30T17:24:37Z

Move content for slurm to an obsolete page

2021-06-27T17:15:22Z

OpenMP / SMP Job: Improve details of SMP invocation with OMP_NUM_THREADS and provide links to additional docs.

← Older revision		Revision as of 17:15, 27 June 2021
Line 140:		Line 140:

	It is very important to properly structure an SMP job to ensure that the requested CPU cores are assigned to the same compute node. The following example requests 4 CPU cores by setting the number of '''ntasks''' to '''1''' and '''cpus-per-tasks''' to '''4'''		It is very important to properly structure an SMP job to ensure that the requested CPU cores are assigned to the same compute node. The following example requests 4 CPU cores by setting the number of '''ntasks''' to '''1''' and '''cpus-per-tasks''' to '''4'''

			For OpenMP you must ensure your OMP_NUM_THREADS environment variable is set and matches then --cpus-per-task requested. This ensures you have requested the 1-core for each thread in your reservation and provides dedicated cores for each thread.

	<pre>		<pre>
			export OMP_NUM_THREADS=4
	srun --partition=short \		srun --partition=short \
	--ntasks=1 \		--ntasks=1 \
	--cpus-per-task=4 \		--cpus-per-task=$OMP_NUM_THREADS \
	--mem-per-cpu=1024 \		--mem-per-cpu=1024 \
	--time=5:00:00 \		--time=5:00:00 \
Line 150:		Line 153:
	--pty /bin/bash		--pty /bin/bash
	</pre>		</pre>

			For additional examples of Slurm with SMP jobs see [https://help.rc.ufl.edu/doc/Sample_SLURM_Scripts#Multi-Threaded_SMP_Job UFL's examples]. For a simple hello world SMP example app see [https://www.geeksforgeeks.org/openmp-hello-world-program/ this code]

	=== Job Dependencies ===		=== Job Dependencies ===

2020-04-16T23:18:25Z

← Older revision		Revision as of 23:18, 16 April 2020
Line 17:		Line 17:
	</pre>		</pre>
	you'll get the manual page for the sbatch command.		you'll get the manual page for the sbatch command.

			Cheatsheets for [https://github.com/wwarriner/slurm_cheatsheets/blob/master/sacct_cheat_sheet.pdf <code>sacct</code>] and [https://github.com/wwarriner/slurm_cheatsheets/blob/master/sbatch_cheat_sheet.pdf <code>sbatch</code>] are available at [https://github.com/wwarriner/slurm_cheatsheets GitHub]. These cheatsheets contain some of the more commonly used flags and parameters for the two commands.

	== Slurm Partitions ==		== Slurm Partitions ==

2020-02-10T21:01:17Z

Slurm Job Types

← Older revision		Revision as of 21:01, 10 February 2020
Line 128:		Line 128:
	* To request an interactive session using a single GPU, say for code development, you can use the following syntax		* To request an interactive session using a single GPU, say for code development, you can use the following syntax
	<pre>		<pre>
	~~sinteracive~~ --partition=pascalnodes --gres=gpu		sinteractive --partition=pascalnodes --gres=gpu
	</pre>		</pre>

Line 148:		Line 148:
	--pty /bin/bash		--pty /bin/bash
	</pre>		</pre>

			=== Job Dependencies ===

			It is also possible to link job scripts using job dependencies. Visit the following git repository for more detailed information and sample scripts: https://gitlab.rc.uab.edu/rc-training-sessions/job-dependency

	== Job Status ==		== Job Status ==

2018-08-22T18:28:53Z

Link to Jupyter Notebook

← Older revision		Revision as of 18:28, 22 August 2018
Line 47:		Line 47:

	== Slurm Job Types ==		== Slurm Job Types ==
			=== Jupyter Job ===
			Cheaha can be used with [[Jupyter]] notebooks.

	=== Batch Job ===		=== Batch Job ===
	'''TODO: ''' provide an explanation of what makes a batch job and why use that vs an interactive job		'''TODO: ''' provide an explanation of what makes a batch job and why use that vs an interactive job

2018-08-15T16:46:29Z

Example Batch Job Script

← Older revision		Revision as of 16:46, 15 August 2018
Line 79:		Line 79:
	#SBATCH --mail-user=YOUR_EMAIL_ADDRESS		#SBATCH --mail-user=YOUR_EMAIL_ADDRESS

			#Set your environment here

			#Run your commands here
	srun hostname		srun hostname
	srun sleep 60		srun sleep 60

2018-08-15T16:45:12Z

Example Batch Job Script

← Older revision		Revision as of 16:45, 15 August 2018
Line 59:		Line 59:
	#		#
	#SBATCH --job-name=test		#SBATCH --job-name=test
	#SBATCH --output=res.~~txt~~		#SBATCH --output=res.out
			#SBATCH --error=res.err
			#
			# Number of tasks needed for this job. Generally, used with MPI jobs
	#SBATCH --ntasks=1		#SBATCH --ntasks=1
	#SBATCH --partition=express		#SBATCH --partition=express
	#		#
	# Time format = HH:MM:SS, DD-HH:MM:SS		# Time format = HH:MM:SS, DD-HH:MM:SS
			#SBATCH --time=10:00
	#		#
	#SBATCH --~~time~~=~~10:00~~		# Number of CPUs allocated to each task.
			#SBATCH --cpus-per-task=1
	#		#
	# Mimimum memory required per allocated CPU in MegaBytes.		# Mimimum memory required per allocated CPU in MegaBytes.
			#SBATCH --mem-per-cpu=100
	#		#
	#~~SBATCH --mem-per-cpu=100~~		# Send mail to the email address when the job fails
	#SBATCH --mail-type=FAIL		#SBATCH --mail-type=FAIL
	#SBATCH --mail-user=YOUR_EMAIL_ADDRESS		#SBATCH --mail-user=YOUR_EMAIL_ADDRESS

2017-10-17T19:37:38Z

Job History

← Older revision		Revision as of 19:37, 17 October 2017
Line 216:		Line 216:
	--format=User,JobID,Jobname,partition,state,time,start,end,elapsed,MaxRss,MaxVMSize,nnodes,ncpus,nodelist		--format=User,JobID,Jobname,partition,state,time,start,end,elapsed,MaxRss,MaxVMSize,nnodes,ncpus,nodelist
	</pre>		</pre>
	<pre style="white-space: pre-wrap; >		<pre style="white-space: pre-wrap;">
	$ rc-sacct --allusers --starttime 2016-08-30		$ rc-sacct --allusers --starttime 2016-08-30

2017-10-17T19:37:16Z

Job History

← Older revision		Revision as of 19:37, 17 October 2017
Line 216:		Line 216:
	--format=User,JobID,Jobname,partition,state,time,start,end,elapsed,MaxRss,MaxVMSize,nnodes,ncpus,nodelist		--format=User,JobID,Jobname,partition,state,time,start,end,elapsed,MaxRss,MaxVMSize,nnodes,ncpus,nodelist
	</pre>		</pre>
	<pre>		<pre style="white-space: pre-wrap; >
	$ rc-sacct --allusers --starttime 2016-08-30		$ rc-sacct --allusers --starttime 2016-08-30

Line 254:		Line 254:

	</pre>		</pre>


	== Slurm Variables ==		== Slurm Variables ==

2017-10-17T19:36:52Z

SSTAT

← Older revision		Revision as of 19:36, 17 October 2017
Line 163:		Line 163:

	'''NOTE: the job parts must be executed using ''srun'' otherwise ''sstat'' will not display useful output'''		'''NOTE: the job parts must be executed using ''srun'' otherwise ''sstat'' will not display useful output'''
	<pre>		<pre style="white-space: pre-wrap;">
	[rcs@login001 ~]$ sstat 256483		[rcs@login001 ~]$ sstat 256483
	JobID MaxVMSize MaxVMSizeNode MaxVMSizeTask AveVMSize MaxRSS MaxRSSNode MaxRSSTask AveRSS MaxPages MaxPagesNode MaxPagesTask AvePages MinCPU MinCPUNode MinCPUTask AveCPU NTasks AveCPUFreq ReqCPUFreqMin ReqCPUFreqMax ReqCPUFreqGov ConsumedEnergy MaxDiskRead MaxDiskReadNode MaxDiskReadTask AveDiskRead MaxDiskWrite MaxDiskWriteNode MaxDiskWriteTask AveDiskWrite		JobID MaxVMSize MaxVMSizeNode MaxVMSizeTask AveVMSize MaxRSS MaxRSSNode MaxRSSTask AveRSS MaxPages MaxPagesNode MaxPagesTask AvePages MinCPU MinCPUNode MinCPUTask AveCPU NTasks AveCPUFreq ReqCPUFreqMin ReqCPUFreqMax ReqCPUFreqGov ConsumedEnergy MaxDiskRead MaxDiskReadNode MaxDiskReadTask AveDiskRead MaxDiskWrite MaxDiskWriteNode MaxDiskWriteTask AveDiskWrite