Running Jobs with Slurm

The Simple Linux Utility for Resource Management (SLURM) is used for submission, scheduling, execution, and monitoring of jobs on Mistral cluster. SLURM is a free open-source resource manager and scheduler. It is a modern, extensible batch system that is widely deployed around the world on clusters of various sizes.

The following topics are treated in this section:

• Slurm Introduction Brief introduction to SLURM commands and ways to allocate resources for job execution and interactive usage of nodes

• Partition and Limits Description of partitions and limits defined on Mistral

• MPI Runtime Settings Recommended settings for environmental variables that influence the MPI library

• Example Batch Scripts Example batch scripts for serial, MPI, OpenMP, and hybrid applications

• Adapting job scripts for mistral phase2 All major steps to transform your existing batch scripts to use phase2 nodes

• Using srun Srun is tasks launcher provided by SLURM. Use it to start MPI and hybrid (MPI/OpenMP) parallel applications on Mistral

• Advanced Slurm Features Submitting and managing of dependency chains and job arrays

• Accounting and Priorities How compute time is accounted and scheduling priorities are calculated

• Job Monitoring How to access node based monitoring data automatically captured for Slurm batch jobs