[TOC title:Table-of-Content]
## slurm
Slurm is a job scheduling system that ensures every submitted job receives the defined hardware allocation. It creates an environment in which jobs can run. It updates the status of jobs and allows users to easily query information about their jobs. It sends email notifications upon request to keep users updated about their job status.
## slurm partitions
The nodes in the cluster are logically separated based on workload and speed. In SLURM, this separation is called 'partitions'. Partitions are designed to help users select the appropriate hardware for their analysis. To avoid long waiting times in the SLURM queue, users should use hardware that fits their analysis requirements.
There are three partitions:
- normal: the default partition with nodes having up to 380 GB of RAM
- bigmem: use this if you need more than 300 GB of RAM
- short: use this for short interactive sessions or for small jobs that run within an hour, max 64 GB of ram.
Use the command 'sinfo' to view the partition details.
## slurm batch script
This is a specific file that users need in order to submit a job. The file defines the necessary resources and how long the job will run for. It also specifies the necessary software and the commands required to perform the specific analysis.
Read more about [slurm batch system](https://servicedesk.surf.nl/wiki/spaces/WIKI/pages/30660218/The+batch+system){:target="_blank"}
## Slurm Queue
After a job is submitted is goes in the so called slurm queue, from their the job will be managed by the slurm job scheduler and update the status of the job. Connected users can see the queue and query specific info about jobs. Read more about [the job scheduler](https://servicedesk.surf.nl/wiki/spaces/WIKI/pages/30660219/The+job+scheduler){:target="_blank"} through this external link.
## module
Users will need to use specific software to run their analyses. The Lmod module works alongside SLURM and allows users to search for and load specific software environments. The software needs to be installed first. Installation is usually carried out by admins, but users can install software locally or via conda for testing purposes.
Request software install by emailing mib.bioinfo@wur.nl, subject "valhalla install [SOFTWARE NAME]