Advanced Filtering Guide

S9S provides powerful filtering capabilities to help you find exactly what you need across thousands of jobs, nodes, and resources. This guide covers all filtering features and syntax.

🔍 Basic Filtering

Quick Filter

Press

in any view to activate quick filter:


# Simple text search
/analysis      # Find items containing "analysis"
/GPU          # Case-insensitive search for "GPU"
/node001      # Find specific node

Clear Filters

```
Esc
```
- Clear current filter and exit search mode
```
:clear
```
- Clear all active filters
```
Ctrl+/
```
- Toggle last filter on/off

🎯 Filter Syntax

Field-Specific Filters

Target specific fields with the

field:value

syntax:


# Job filters
/user:alice              # Jobs by user alice
/name:simulation        # Jobs with "simulation" in name
/jobid:12345           # Specific job ID
/state:RUNNING         # Jobs in RUNNING state
/partition:gpu         # Jobs in GPU partition

# Node filters
/node:node001          # Specific node
/state:idle           # Idle nodes
/features:gpu         # Nodes with GPU feature
/memory:>128GB       # Nodes with >128GB RAM

Operators

S9S supports various comparison operators:

Operator	Description	Example
`:`	Equals	`/state:RUNNING`
`!`	Not equals	`/state:!FAILED`
`>`	Greater than	`/nodes:>4`
`<`	Less than	`/runtime:<1h`
`>=`	Greater or equal	`/priority:>=1000`
`<=`	Less or equal	`/memory:<=64GB`
`~`	Regex match	`/name:~"test.*2023"`
`*`	Wildcard	`/name:analysis*`
`..`	Range	`/nodes:2..8`

📊 Advanced Filtering

Compound Filters

Combine multiple filters with spaces (AND logic):


# Jobs by alice in GPU partition
/user:alice partition:gpu

# Running jobs with more than 4 nodes
/state:RUNNING nodes:>4

# Failed jobs submitted today
/state:FAILED submitted:today

OR Logic

Use comma-separated values for OR logic:


# Jobs in RUNNING or PENDING state
/state:RUNNING,PENDING

# Jobs in gpu or cpu partition  
/partition:gpu,cpu

# Jobs by alice or bob
/user:alice,bob

Complex Queries

Combine AND and OR logic:


# Alice or Bob's GPU jobs
/user:alice,bob partition:gpu state:RUNNING

# High priority pending jobs in specific partitions
/state:PENDING priority:>1000 partition:gpu,highmem

🕐 Time-Based Filters

Relative Time

Use human-readable time expressions:


# Submitted time
/submitted:<1h         # Less than 1 hour ago
/submitted:>1d        # More than 1 day ago
/submitted:today      # Submitted today
/submitted:yesterday  # Submitted yesterday
/submitted:thisweek   # This week
/submitted:lastweek   # Last week

# Runtime
/runtime:>30m         # Running more than 30 minutes
/runtime:<2h         # Running less than 2 hours
/runtime:1h-3h      # Between 1 and 3 hours

Absolute Time

Use specific dates and times:


# ISO format
/submitted:>2023-12-01
/started:<2023-12-25T18:00:00

# Date shortcuts
/submitted:2023-12-01..2023-12-31
/ended:yesterday..today

🔢 Numeric Filters

Resource Filters

Filter by resource usage:


# Node count
/nodes:4              # Exactly 4 nodes
/nodes:>8            # More than 8 nodes
/nodes:2..16        # Between 2 and 16 nodes

# Memory (supports units)
/memory:>32GB        # More than 32GB
/memory:64GB..128GB  # Between 64GB and 128GB
/memory:<=1TB       # Up to 1TB

# CPUs
/cpus:>48           # More than 48 CPUs
/cpus:24,48,96     # Specific CPU counts

# GPUs
/gpus:>0           # Has GPUs
/gpus:8            # Exactly 8 GPUs

Priority and QoS


# Priority
/priority:>1000      # High priority jobs
/priority:0..500    # Low to medium priority

# QoS
/qos:normal         # Normal QoS
/qos:high,critical  # High or critical QoS

🎨 Regular Expressions

Use regex for complex pattern matching:


# Enable regex with ~ operator
/name:~"analysis_\d{4}"      # analysis_0001, analysis_0002, etc.
/user:~"(alice|bob)_.*"      # alice_* or bob_* users
/output:~"error|warning"      # Find errors or warnings
/script:~"python.*\.py$"     # Python scripts

# Case-insensitive regex
/name:~"(?i)GPU"             # Matches gpu, GPU, Gpu, etc.

# Negative lookahead
/name:~"^(?!test).*"         # Not starting with "test"

🏷️ State Filters

Job States


# Single state
/state:RUNNING
/state:PENDING
/state:COMPLETED
/state:FAILED

# Multiple states
/state:RUNNING,PENDING
/state:!COMPLETED        # Not completed

# State groups (S9S shortcuts)
/state:active           # RUNNING,PENDING
/state:ended           # COMPLETED,FAILED,CANCELLED
/state:problem         # FAILED,TIMEOUT,NODE_FAIL

Node States


# Basic states
/state:idle
/state:allocated
/state:down
/state:drain

# Compound states
/state:idle+drain      # Idle and draining
/state:allocated+drain # Allocated but draining

# State shortcuts
/state:available       # idle,mixed
/state:unusable       # down,drain,maint

🔖 Saved Filters

Creating Saved Filters

Save frequently used filters:


# Save current filter
:filter save my-jobs "/user:${USER} state:active"

# Save with description
:filter save gpu-pending "/partition:gpu state:PENDING" --desc "Pending GPU jobs"

# Save complex filter
:filter save inefficient "/state:RUNNING efficiency:<0.5 runtime:>1h"

Using Saved Filters


# List saved filters
:filter list

# Load saved filter
:filter load my-jobs
# Or use shortcut
/~my-jobs

# Edit saved filter
:filter edit gpu-pending

# Delete saved filter
:filter delete old-filter

Filter Presets

S9S includes built-in filter presets:

Preset	Description	Filter
`~active`	Active jobs	`state:RUNNING,PENDING`
`~mine`	My jobs	`user:${USER}`
`~recent`	Recent jobs	`submitted:<1d`
`~failed`	Failed jobs	`state:FAILED,TIMEOUT`
`~gpu`	GPU jobs	`gpus:>0`
`~long`	Long jobs	`runtime:>1d`
`~today`	Today's jobs	`submitted:today`

🔄 Dynamic Filters

Auto-Refresh Filters

Filters that update automatically:


# Jobs submitted in last hour (updates)
/submitted:<1h

# Currently running jobs
/state:RUNNING elapsed:>0

# Recently completed
/state:COMPLETED ended:<10m

Context-Aware Filters

Filters that adapt to current view:


# In Jobs view
/efficiency:<0.7      # Low efficiency jobs

# In Nodes view  
/load:>0.8           # High load nodes

# In Users view
/active:true         # Users with running jobs

📝 Filter Examples

Common Use Cases

Find Stuck Jobs


/state:PENDING submitted:>1h reason:!Resources

Inefficient GPU Jobs


/partition:gpu state:RUNNING gpu_util:<50%

Failed Jobs Today


/state:FAILED ended:today user:${USER}

High Memory Jobs


/memory:>500GB state:RUNNING,PENDING

Jobs Near Time Limit


/state:RUNNING time_left:<30m

Power User Filters

Complex Resource Query


/nodes:>16 cpus:>512 memory:>1TB partition:large state:PENDING

Multi-User Team Filter


/user:~"^(alice|bob|charlie)" project:ml_research state:!COMPLETED

Performance Analysis


/state:RUNNING runtime:>2h efficiency:<0.5 partition:!debug

⚡ Performance Tips

Efficient Filtering

Use indexes: Filter by indexed fields first (JobID, User, State)
Narrow scope: Start with restrictive filters, then broaden
Save complex filters: Reuse instead of retyping
Use shortcuts: Learn preset filters for common queries

Filter Optimization


# Good: Indexed field first
/state:RUNNING user:alice partition:gpu

# Less efficient: Unindexed field first
/name:analysis state:RUNNING user:alice

🎯 Filter Shortcuts

Keyboard Shortcuts

Key	Action	Description
`/`	Start filter	Enter filter mode
`Tab`	Autocomplete	Complete filter fields
`↑/↓`	History	Browse filter history
`Ctrl+/`	Toggle	Toggle last filter
`Alt+/`	Presets	Show filter presets
`Esc`	Clear	Clear current filter

Command Mode Filters


# Apply filter via command
:filter "state:RUNNING nodes:>4"

# Chain filters
:filter add "user:alice"
:filter add "partition:gpu"

# Remove specific filter
:filter remove "user:alice"

🔧 Troubleshooting Filters

Common Issues

No Results

Check filter syntax
Verify field names
Try broader criteria
Check for typos

Too Many Results

Add more specific criteria
Use compound filters
Filter by state first
Add time constraints

Slow Filters

Avoid regex on large datasets
Use indexed fields first
Limit time range
Consider saved filters

🚀 Next Steps

Practice filters in Mock Mode
Set up Saved Filters
Learn Batch Operations with filters
Explore Export Options for filtered data