PBS to Slurm | zh

--- title: PBS to Slurm | zh tags: Guide, TWNIA3, TW GA: --- {%hackmd @docsharedstyle/default %} # Taiwania 1 與 Taiwania 3 排程系統指令對照 Taiwania 1 與 Taiwania 3 所使用的排程系統不同，Taiwania 1 為PBS pro，Taiwania 3 為 SLURM。原 Taiwania 1 用戶可參考以下相關常用指令對照表以便理解 SLURM 的工作腳本 (job script) 寫法。 ## PBS to Slurm ### 任務常用指令 | Function | PBS | Slurm | | :--------: | :--------: | :--------: | | 任務提交 |qsub [script_file]|sbatch [script_file]| | 任務刪除 |qdel [job_id]|scancel [job_id]| |(job_id)任務狀態 |qstat [job_id]|squeue | |(user_name)任務狀態 |qstat -u [user_name]|squeue -u [user_name]| |任務釋出 |qrls [job_id]|scontrol release [job_id]| |任務集狀態 |qstat -Q|squeue| |節點列表|pbsnodes -l|sinfo -N / scontrol show nodes| |群集狀態|qstat -a|sinfo|  ### Job Script 指令參數對照 | Script 指令 | PBS | Slurm | | :--------: | :--------: |:--------:| |Script Directive |#PBS|#SBATCH| |Queue/Partition|-q [name]|-p [name] / --partition=[name] | |Node Count|-l nodes=[count]|-N [count] / --nodes=[count] | |Total Task Count|-l ppn=[count] / -l mppwidth=[PE_count]|-n [ntasks, total processes] / --ntasks=[ntasks]| |Wall Clock Limit|-l walltime=[hh:mm:ss]|-t [days-hh:mm:ss] / --time=[days-hh:mm:ss]| |Standard Output File|-o [file_name]|-o [file_name] / --output=[file_name]| |Standard Error File|-e [file_name]|-e [file_name] / --error=[file_name]| | Combine stdout/err | -j oe (both to stdout) / -j eo (both to stderr) | (use -o without -e) | | Copy Environment|-V|--export=[ALL, NONE, variables]| | Event Notification|-m abe| --mail-type=[events]| | Email Address|-M [address]| --mail-user=[address]| | Job Name|-N [name]|-J [name] / --job-name=[name]| | Job Restart|-r [y, n]|--requeue / --no-requeue| | Memory Size |-l mem=[MB] |--mem=[mem][M / G / T] / --mem-per-cpu=[mem][M / G / T]| |Accounts to charge|-P OR -W group_list=[account]|-A [account] /　--account=[account]| |Tasks Per Node|-l mppnppn [PEs_per_node]| --tasks-per-node=[count]| |CPUs Per Task| | --cpus-per-task=[count]| |Job Dependency|-d [job_id]|-d [state:job_id] / --depend=[state:job_id]| |Job Arrays|-t [array_spec]|-a [array_spec] / --array=[array_spec]| |Generic Resources|-l other=[resource_spec]| --gres=[resource_spec]| --- ### Taiwania 1 VS. Taiwania 3 job script 範例 :::success <i class="fa fa-paperclip fa-20" aria-hidden="true"></i> **說明：** **台灣杉一號每個節點40核心，台灣杉三號每個節點56核心。** 以下範例為協助台灣杉一號用戶轉換為台灣杉三號時供對照參考，請依實際需求與設備規格做適當調整。 ::: #### Taiwania 1 ``` ############################################### # Intel MPI job script example # ############################################### #!/bin/bash #PBS -P TRI107693 #PBS -N sample_job #PBS -l select=2:ncpus=40:mpiprocs=40 #PBS -l walltime=00:30:00 #PBS -q ctest #PBS -o jobresult.out #PBS -e jobresult.err module load intel/2018_u1 cd ${PBS_O_WORKDIR:-"."} mpirun ./myprogram #本例為使用2個節點，每個節點使用40核心、執行40個mpiprocesses，共計使用80個核心、執行80個mpiprocesses ``` #### Taiwania 3 ``` #!/bin/bash #SBATCH --account=TRI107693 # (-A) Account/project number #SBATCH --job-name=sample_job # (-J) Job name #SBATCH --partition=ctest # (-p) Specific slurm partition #SBATCH --nodes=2 # (-N) Maximum number of nodes to be allocated #SBATCH --ntasks-per-node=40 # Maximum number of tasks on each node #SBATCH --time=00:30:00 # (-t) Wall time limit (days-hrs:min:sec) #SBATCH --output=%j.log # (-o) Path to the standard output and error files relative to the working directory #SBATCH --error=%j.err # (-e) Path to the standard error ouput #SBATCH --mail-type=END,FAIL # Mail events (NONE, BEGIN, END, FAIL, ALL) #SBATCH --mail-user=user@mybox.mail # Where to send mail. Set this to your email address module load compiler/intel/2020u4 IntelMPI/2020 mpirun /path/to/ypur_program or mpiexec.hydra -n $SLURM_NTASKS /path/to/ypur_program ``` 亦可參考以下寫法： ``` #!/bin/bash #SBATCH -A TRI107693 #SBATCH -J sample_job #SBATCH -p ctest #SBATCH -N 2 #SBATCH --ntasks-per-node=40 #SBATCH -t 00:30:00 #SBATCH -o %j.log #SBATCH -e %j.err #SBATCH --mail-type=END,FAIL #SBATCH --mail-user=user@mybox.mail module load compiler/intel/2020u4 IntelMPI/2020 mpirun /path/to/ypur_program or mpiexec.hydra -n $SLURM_NTASKS /path/to/ypur_program ``` ---