Monitor Slurm Jobs

Monitor currently running Slurm jobs and inspect their logs for errors or issues.

Usage

/monitor-slurm

Features

• •Lists all running Slurm jobs for your account
• •If no running jobs, inspects the k most recently added job IDs from the logs directory and returns a big picture summary
• •Automatically finds matching .out and .err files in ./<project_name>/logs/
• •Displays last N lines of each log file for quick diagnostics
• •Shows job status summary

Options

• •--tail N: Show last N lines of log files (default: 20)
• •--full: Show complete log file contents instead of tail
• •--verbose: Show detailed job information

Examples

# Basic monitoring
/monitor-slurm

# Show more log lines
/monitor-slurm --tail 50

# Show full log contents
/monitor-slurm --full

# Verbose output with all job details
/monitor-slurm --verbose

How it works

1. •Queries active Slurm jobs using squeue
2. •If running jobs found:
   • •For each job, searches ./<project_name>/logs/ for matching log files
   • •Matches files by job ID pattern: <job_id>_*.out and <job_id>_*.err
   • •Displays log contents to help identify issues
   • •Provides job status summary
3. •If no running jobs:
   • •Takes the k most recently added job IDs from the logs directory
   • •Inspects their log files
   • •Returns a big picture summary of recent job history

Log File Matching

The skill expects log files named by job ID as:

• •<job_id>_<rest_of_name>.out
• •<job_id>_<rest_of_name>.err

For example, if your job ID is 12345, it will find:

• •12345_*.out
• •12345_*.err