slurmconf

Slurmconf, commonly referred to by its conventional filename slurm.conf, is the main configuration file used by the Slurm Workload Manager. It is read by both the Slurm controller (slurmctld) and the Slurm daemon on compute nodes (slurmd) to determine how the cluster should operate, how resources are allocated, and what policies govern scheduling, authentication, and accounting.

The file is a plain text document, typically located in /etc/slurm/slurm.conf or a closely related path, but exact locations can vary by installation. Administrators may generate it from templates, use a configuration tool provided by Slurm, or edit it directly. Slurm also supports including additional configuration fragments from other files via Include directives, which helps organize large clusters.

Slurm.conf is structured with global settings and resource definitions. Global parameters cover behavior and system-wide policies, such as the authentication method (AuthType), the scheduler (SchedulerType), the user under which Slurm runs (SlurmUser), ports, and location of state and logging data. The resource definitions consist of NodeName lines, which describe compute nodes (for example, CPUs, memory, sockets, cores per socket, and current state), and PartitionName lines, which define partitions or queues, including which nodes are members, default status, maximum job time, and current state.

Changes to slurm.conf typically require reloading or restarting Slurm daemons. Administrators can validate configuration with tools like scontrol show config or by issuing a reconfigure command. Properly configured slurm.conf is essential for correct cluster operation, effective resource usage, and predictable job scheduling.