Understanding how your job uses CPU, memory, and other resources can help you optimize performance and avoid wasted allocations. On the MocsHPC cluster, you can use Jobstats to view detailed usage metrics for your jobs.
Option 1: Using the Jobstats Command Line Tool
After your job has started (or completed), you can view its resource usage from the command line.
Step 1: Run the jobstats command
jobstats <jobid>
Replace <jobid> with your Slurm job ID number.
Step 2: Review the output
The command will display a summary of your job’s resource usage, such as:
- CPU usage
- Memory usage
- Runtime
- Node information
It will also include a URL to interactive charts.
Step 3: Open the charts
Copy and paste the provided URL into your browser to view detailed graphs, including:
- CPU utilization over time
- Memory usage over time
- GPU usage (if applicable)
- Storage performance
Option 2: Using Open OnDemand
If you are using Open OnDemand, Jobstats information is integrated directly into the interface.
While Your Job Is Running
- Navigate to your interactive app dashboard.
- Locate your running session.
- In the app card, you will see a link to Jobstats charts.
- You can also view stats in the top menu at "Jobs" -> "Active Jobs"
Click the link to monitor your job in real time.
After Your Job Completes
Once the job finishes:
- The app card will display the jobstats command output, similar to what you would see in the terminal.
- This includes a link to the charts for post-run analysis.
Tips for Interpreting Jobstats
- Low CPU usage: Your job may not be fully utilizing allocated cores.
- High memory usage: You may need to request more memory to avoid failures.
- Consistently low utilization: Consider requesting fewer resources to reduce queue time.
When to Use Jobstats
Use Jobstats when you want to:
- Debug performance issues
- Tune resource requests for future jobs
- Verify that your job is using GPUs or CPUs as expected
Additional Support
Open an IT Helpdesk request ticket.
Send an email to ITHelp@utc.edu.