Dashboard View
The Dashboard provides a real-time overview of your SLURM cluster with comprehensive metrics, health indicators, and quick navigation.
Dashboard showing cluster metrics, job distribution, and node status
Overview
The Dashboard provides a cluster overview accessible via key 8, displaying six information panels organized in a responsive layout:
- Cluster Overview - System information and health status
- Jobs Summary - Job queue and state distribution
- Nodes Summary - Node availability and resource usage
- Partition Status - Top partitions by size
- Alerts & Issues - System health warnings
- Performance Trends - Historical throughput and efficiency
Display Panels
Cluster Overview (Top Left)
Displays essential cluster information:
- Cluster name, version, and endpoint
- Overall health status (EXCELLENT/GOOD/FAIR/POOR/CRITICAL)
- CPU and Memory usage percentages
- Visual utilization bars
Health scoring is calculated from:
- Down nodes (2 points per %)
- Failed jobs (1 point per %)
- High CPU/Memory utilization (up to 10 points each for >95%)
Jobs Summary (Top Center)
Shows current job queue status:
- Total job count
- Job state breakdown:
- Running - Currently executing jobs
- Pending - Queued jobs waiting for resources
- Completed - Successfully finished jobs
- Failed - Jobs that ended with errors
- Canceled - User-canceled jobs
- Job state distribution visualization
- Average wait time for pending jobs
Nodes Summary (Top Right)
Provides node availability metrics:
- Total node count
- Node state counts:
- Idle - Available for work
- Allocated - Fully assigned
- Mixed - Partially allocated
- Down - Unavailable
- Drain - Being drained
- CPU and Memory utilization percentages
- Node availability percentage
Partition Status (Middle Left)
Lists the top 8 partitions by size:
- Partition name
- Node count
- CPU count
- Current state
- Sorted by node count (largest first)
Alerts & Issues (Middle Right)
Displays system alerts when issues are detected:
- Down nodes warning
- High memory utilization (>90%)
- High CPU utilization (>90%)
- Long waiting jobs (>24 hours)
- Failed jobs (>10)
Shows "No issues detected" with green checkmark when system is healthy.
Performance Trends (Bottom)
Displays cluster performance visualizations:
- Resource efficiency bar (average of CPU and Memory utilization)
- System health score with visual bar
These values are computed from real cluster metrics collected during the session.
Actions & Shortcuts
Layout
| Key | Action |
|---|---|
| L | Switch layout (Default / Monitoring) |
Analytics
| Key | Action |
|---|---|
| A | Show Advanced Analytics modal |
| H | Show Health Check modal |
Data Management
| Key | Action |
|---|---|
| R | Manual refresh all dashboard data |
Advanced Analytics Modal
Press A to open the Advanced Analytics modal with comprehensive cluster analysis:
Resource Efficiency
- CPU efficiency percentage and assessment
- Memory efficiency percentage and assessment
- Overall resource efficiency rating
Job Analysis
- Job state distribution
- Average wait times by state
- Queue depth analysis
Node Analysis
- Node state distribution
- Node utilization metrics
- Resource availability trends
Recommendations
The system applies rule-based heuristics to current metrics and provides:
- Performance optimization suggestions
- Resource allocation recommendations
- Potential issue warnings
Modal Shortcuts:
- R - Refresh analytics
- ESC - Close modal
Health Check Modal
Press H to open detailed health check report:
Overall Health Score
- Numeric score (0-100)
- Color-coded health rating
- Status summary
Component Health Checks
Nodes
- Down node count and percentage
- Draining node count
- Impact assessment
Job Queue
- Failed job count
- Long-running stuck jobs
- Queue health status
Resource Utilization
- CPU usage vs. critical threshold
- Memory usage vs. critical threshold
- Utilization warnings
Partitions
- Operational partition count
- Down partition count
- Partition availability
Each component shows:
- Status indicator (✓/⚠/✗)
- Detailed metrics
- Threshold information
Modal Shortcuts:
- R - Refresh health check
- ESC - Close modal
Auto-Refresh
The Dashboard loads data on initial display but does not automatically refresh afterward. Press R for a manual refresh to update the dashboard data.
Visual Design
The Dashboard uses a 6-panel color system for status indicators:
- Green - Healthy, good status
- Red - Critical issues, failures
- Yellow - Warnings, attention needed
- Orange - Moderate issues
- Cyan - Informational
- Teal - Headers and accents
Technical Details
Data Fetching
The Dashboard fetches data concurrently from multiple sources:
- Cluster information and version
- Current metrics (CPU/Memory)
- Job list with all states
- Node list with states
- Partition information
This parallel fetching ensures quick dashboard updates even on large clusters.
Layout Adaptation
The Dashboard uses flexible layout proportions to adapt to different terminal sizes:
- Minimum recommended width: 120 columns
- Minimum recommended height: 30 rows
- Panels resize proportionally
- Text truncates gracefully
Tips
- Use the Dashboard as your cluster monitoring hub
- Press A regularly to review recommendations
- Check H for detailed health diagnostics
- Press L to switch between the default dashboard and the monitoring layout
- Press R to manually refresh the dashboard data when you want the latest metrics