S9Sv1.0

Views Overview

S9S organizes cluster information into specialized views, each optimized for specific tasks. This guide provides a comprehensive overview of all available views.

📊 Available Views

1. Jobs View

Purpose: Monitor and manage SLURM jobs
Default Key: 

1

The Jobs view is the primary interface for job management:

  • Real-time Updates: Live job status updates
  • Detailed Information: Job ID, name, user, state, runtime, nodes
  • Quick Actions: Cancel, hold, release, view output
  • Advanced Filtering: Filter by state, user, partition, time
  • Batch Operations: Manage multiple jobs simultaneously

Key features:

  • Color-coded job states (RUNNING=green, PENDING=yellow, FAILED=red)
  • Expandable job details with resource usage
  • Direct access to job scripts and output files
  • Job dependency visualization
  • Performance metrics per job

2. Nodes View

Purpose: Monitor compute nodes and system health
Default Key: 

2

Comprehensive node management interface:

  • Node Status: Real-time node states and health
  • Resource Usage: CPU, memory, GPU utilization
  • Maintenance: Drain/resume operations
  • SSH Access: Direct terminal access to nodes

Features:

  • Visual load indicators
  • Temperature monitoring (if available)
  • Node history and statistics
  • Allocated vs. available resources
  • Network status and interconnect health

3. Partitions View

Purpose: View and manage SLURM partitions
Default Key: 

3

Partition overview and management:

  • Partition Status: State, limits, and configuration
  • Resource Summary: Total/allocated/available resources
  • Job Distribution: Jobs per partition
  • Access Control: Allowed users and accounts

Key information:

  • Node count and allocation
  • Time limits (min/max/default)
  • Priority and preemption settings
  • QoS associations
  • Partition-specific features (GPU, high-mem, etc.)

4. Users View

Purpose: User activity and resource usage
Default Key: 

4

Monitor user activity across the cluster:

  • Active Users: Currently running/pending jobs
  • Resource Usage: CPU hours, memory, storage
  • Fair Share: Priority and fairshare statistics
  • Job History: Recent job submissions

Features:

  • User quotas and limits
  • Group associations
  • Resource consumption trends
  • Efficiency metrics
  • Account associations

5. Accounts View

Purpose: Account management and billing
Default Key: 

5

Manage SLURM accounts and hierarchies:

  • Account Tree: Hierarchical account structure
  • Resource Limits: Account-specific limits
  • Usage Tracking: Resource consumption by account
  • User Associations: Users per account

Information displayed:

  • Account hierarchy visualization
  • Allocation and usage
  • Fair share statistics
  • QoS assignments
  • Billing information (if configured)

6. QoS View

Purpose: Quality of Service policies
Default Key: 

6

Manage QoS configurations:

  • QoS List: All defined QoS levels
  • Limits: Resource limits per QoS
  • Priority: QoS priority levels
  • Associations: Users/accounts using each QoS

Key features:

  • Preemption policies
  • Time limits
  • Resource limits (CPU, GPU, memory)
  • Priority modifiers
  • Usage statistics per QoS

7. Reservations View

Purpose: System reservations management
Default Key: 

7

View and manage system reservations:

  • Active Reservations: Current and upcoming
  • Resource Allocation: Reserved nodes/resources
  • Time Windows: Start and end times
  • Access Lists: Authorized users/accounts

Features:

  • Reservation timeline
  • Resource overlap detection
  • Utilization statistics
  • Recurring reservation patterns
  • Maintenance windows

8. Dashboard View

Purpose: Cluster overview and metrics
Default Key: 

8

High-level cluster status and metrics:

  • Cluster Health: Overall system status
  • Resource Utilization: CPU, memory, GPU usage
  • Job Statistics: Running, pending, completed
  • Performance Metrics: Throughput and efficiency

Widgets include:

  • Utilization gauges
  • Job state distribution
  • Queue wait times
  • Top users by usage
  • Recent events and alerts
  • Cluster efficiency trends

9. Health View

Purpose: System health monitoring
Default Key: 

9

Comprehensive health monitoring:

  • System Alerts: Critical issues and warnings
  • Component Status: SLURM daemons, database, filesystem
  • Performance Issues: Bottlenecks and degradation
  • Maintenance Events: Scheduled and unscheduled

Monitoring includes:

  • Service availability
  • Response times
  • Error rates
  • Resource exhaustion warnings
  • Configuration issues

🎯 Specialized Views

Performance View

Access: 

:view performance

Detailed performance analytics:

  • Job efficiency metrics
  • Resource utilization trends
  • Bottleneck analysis
  • Historical comparisons
  • Optimization recommendations

Stream Monitor

Access: 

:view stream

Real-time log streaming:

  • Job output monitoring
  • System log aggregation
  • Error detection
  • Pattern matching
  • Alert triggers

SSH Terminal

Access: 

s
in Nodes view

Integrated SSH client:

  • Direct node access
  • Session management
  • File transfer support
  • Terminal multiplexing
  • Session recording

Batch Operations

Access: 

b
in supported views

Bulk action interface:

  • Multi-select operations
  • Batch job management
  • Bulk configuration changes
  • Mass notifications
  • Scheduled actions

Job Submission Wizard

Access: 

s
in Jobs view

Guided job submission:

  • Template selection
  • Resource calculation
  • Dependency setup
  • Script validation
  • Submission preview

🎨 View Customization

Column Configuration

Customize visible columns per view:

# In command mode
:set columns JobID,Name,User,State,Time,Nodes

# Save as default
:save-columns

Sorting Options

Sort by any column:

  • Click column headers
  • Use 
    :sort <column>
    command
  • o
    key for sort menu

View Layouts

Save and load custom layouts:

:save-layout my-layout
:load-layout my-layout
:delete-layout old-layout

Split Views

Work with multiple views simultaneously:

:split              # Horizontal split
:vsplit             # Vertical split
:close              # Close current pane

📊 Data Export

Export data from any view:

Export Formats

  • CSV: Spreadsheet compatible
  • JSON: Structured data
  • Markdown: Documentation ready
  • HTML: Web reports
  • XML: System integration

Export Commands

:export csv --file=jobs.csv
:export json --filter="state:RUNNING"
:export markdown --columns="JobID,Name,State"

🔧 View Settings

Auto-refresh

Configure refresh rates per view:

:set refresh 5s      # 5-second refresh
:set refresh 0       # Disable auto-refresh
:auto-refresh on/off # Toggle

Filters

Save frequently used filters:

:filter save "gpu-jobs" "partition:gpu state:RUNNING"
:filter load "gpu-jobs"
:filter list
:filter delete "old-filter"

Themes

Customize view appearance:

:theme dark
:theme light
:theme high-contrast
:theme custom ~/.s9s/themes/my-theme.yaml

💡 Tips for Each View

Jobs View Tips

  • Use 
    o
    for quick output viewing
  • Shift+O
    opens output in external editor
  • Group by user/partition with 
    :group-by
  • Save job templates with 
    :save-template

Nodes View Tips

  • g
    shows GPU details if available
  • Temperature view with 
    t
  • Network topology with 
    :topology
  • Historical graphs with 
    h

Dashboard Tips

  • Customize widgets with 
    c
  • Drag widgets to rearrange
  • Export dashboards with 
    :export-dashboard
  • Create alerts with 
    :alert-on

🚀 Next Steps