SlurmRunner¶
Defined in scheduler.runners.slurm_runner
A runner that submits jobs to a Slurm cluster. This runner creates temporary job scripts and submits them to Slurm. It can handle different job types: - Function: Serializes and runs Python functions - Script: Executes scripts directly - Container: Runs containers using Singularity
Inherits from: BaseRunner
Class Definition¶
class SlurmRunner(self, partition: <class 'str'> = batch, time_limit: <class 'str'> = 01:00:00, memory: <class 'str'> = 4G, cpus_per_task: <class 'int'> = 1, config: Dict[str, Any] = None):
"""
Initialize a new SlurmRunner.
**Args:**
* **partition**: Slurm partition to submit jobs to
* **time_limit**: Time limit for jobs (HH:MM:SS)
* **memory**: Memory to allocate per job
* **cpus_per_task**: Number of CPUs to allocate per job
* **config**: Additional configuration options:
* **modules**: List of modules to load (default: ['python'])
* **singularity_path**: Path to singularity executable (default: 'singularity')
* **job_dir**: Directory to store job files (default: ~/slurm_jobs)
"""
Methods¶
| Method | Description |
|---|---|
cancel_job |
Cancel a job. |
check_job_status |
Check the status of a job and update its state. |
run_job |
Submit a job to Slurm. |
Method Details¶
cancel_job¶
Cancel a job. Args: * job: The job to cancel
check_job_status¶
Check the status of a job and update its state. Args: * job: The job to check
run_job¶
Submit a job to Slurm. Args: * job: The job to run