Users View

The Users view displays user account information, administrative levels, resource limits, and account associations.

Users view showing user accounts, admin levels, and resource limits

Overview

The Users view provides a comprehensive list of all SLURM users with their permissions, resource limits, and account memberships. Use this view to manage user access and understand resource allocation policies.

Table Columns

Admin Levels

Admin levels are color-coded for quick identification:

Admin Level Capabilities

Administrator:

• Modify cluster configuration
• Create/delete users and accounts
• Drain/resume nodes
• Cancel any user's jobs
• Modify QoS policies
• View all cluster data

Operator:

• Monitor cluster status
• Drain/resume nodes
• View all jobs
• Limited job control
• No configuration changes

Regular:

• Submit jobs
• View own jobs
• Cancel own jobs
• View cluster status

Resource Limits

Max Jobs

Maximum number of concurrent jobs a user can have in the system (running + pending).

• Unlimited: Shown as "∞" or "unlimited"
• Limited: Specific number (e.g., "100")

When limit is reached, new job submissions are rejected until running jobs complete.

Max Nodes

Maximum number of nodes a single job can request.

• Prevents users from monopolizing large partition resources
• Can be set per-partition
• "Unlimited" means subject to partition limits only

Max CPUs

Maximum number of CPUs across all concurrent jobs.

• Total CPU allocation limit
• Affects both running and pending jobs
• Helps ensure fair resource sharing

Example:

Copy
User: alice
Max CPUs: 1000
Current usage: 750 CPUs (3 jobs)
Available: 250 CPUs for new jobs

User Actions

View User Details

Shortcut:

Enter

Shows comprehensive user information:

Account Information:

• Username and UID
• Default account
• Associated accounts with priorities
• Administrative level

QoS Settings:

• Default QoS
• Allowed QoS levels
• QoS priorities

Resource Limits:

• Maximum jobs (running + pending)
• Maximum nodes per job
• Maximum CPUs total
• Maximum wall time
• Maximum memory

Current Usage:

• Active jobs count
• Current CPU usage
• Current node usage
• Current memory usage

Account Associations:

• List of accounts user can submit to
• Default account designation
• Priority levels per account

Filter Users

Simple Filter

Shortcut:

/

Filter by:

• Username
• Default account
• Admin level
• QoS
• Any displayed field

Advanced Filter

Shortcut:

F3

Expression-based filtering:

Copy
admin:Administrator
account:research
qos:high
maxjobs:>50

Supported fields:

• Copy

  user

  - Username
• Copy

  uid

  - User ID
• Copy

  account

  - Default account name
• Copy

  admin

  - Admin level (Administrator/Operator/Regular)
• Copy

  qos

  - Default QoS
• Copy

  maxjobs

  - Max jobs limit (supports >, <, >=, <=)
• Copy

  maxnodes

  - Max nodes limit (supports comparison)
• Copy

  maxcpus

  - Max CPUs limit (supports comparison)

Admin Filter Toggle

Shortcut:

a/A

Toggles between:

• All users: Shows everyone (default)
• Admins only: Shows only Administrators and Operators

Useful for quickly finding users with elevated privileges.

Global Search

Shortcut:

Ctrl+F

Search across all cluster resources.

Sorting

Sort users by clicking column headers or using number keys.

Useful sorting:

• By admin level (find administrators)
• By default account (group by research team)
• By max CPUs (identify power users)
• By username (alphabetical)

Press

1-9

Keyboard Shortcuts Reference

User Operations

Enter

Filtering

/

F3

Ctrl+F

a/A

ESC

Data Management

R

1-9

User Details Example

When viewing user details (

Enter

Copy
User: alice
UID: 1234
Admin Level: Operator

Default Account: ml-team
Default QoS: normal

Resource Limits:
  Max Jobs: 50
  Max Nodes: 16 per job
  Max CPUs: 1000 total
  Max Wall Time: 7-00:00:00 (7 days)
  Max Memory: 2TB

Current Usage:
  Active Jobs: 8
  CPUs Used: 384/1000 (38%)
  Nodes Used: 12/16 (75%)

Associated Accounts:
  ml-team (Default, Priority: 100)
  research (Priority: 50)
  shared (Priority: 10)

Allowed QoS Levels:
  normal (Default)
  high
  low

Understanding Resource Limits

Per-User vs. Per-Job Limits

Per-User Limits (enforced across all jobs):

• Max Jobs - total concurrent jobs
• Max CPUs - total CPU allocation
• Max Nodes - cumulative node usage

Per-Job Limits (enforced per individual job):

• Max Nodes per Job - single job cannot exceed
• Max CPUs per Job - single job cannot exceed
• Max Memory per Job - single job cannot exceed

Example scenario:

Copy
User settings:
  Max Jobs: 10
  Max CPUs: 1000
  Max Nodes per Job: 50

Valid: 10 jobs × 100 CPUs each = 1000 CPUs total ✓
Invalid: 5 jobs × 250 CPUs each = 1250 CPUs total ✗
Invalid: 1 job × 60 nodes = exceeds per-job limit ✗

Unlimited Settings

When a limit shows "∞" or "unlimited":

• No user-specific limit set
• Still subject to partition limits
• Still subject to QoS limits
• Still subject to cluster-wide policies

Account Associations

Users can belong to multiple accounts:

Default Account:

• Used when no account is specified in job submission
• Shown in Default Account column
• Can be changed by user or admin

Secondary Accounts:

• User can submit jobs to any associated account
• Must specify with
  Copy

  --account=name

  flag
• Useful for billing different projects separately

Priority Levels:

• Each association has a priority
• Higher priority = better scheduling
• Affects fair-share calculation

Admin-Only View

Press

a/A

Shows:

• Administrators (red)
• Operators (yellow)

Hides:

• Regular users

Use cases:

• Find who can help with admin tasks
• Audit admin access
• Contact operators for assistance

Tips

• Check resource limits: Before requesting help with "job won't submit" issues
• Review account associations: Users might not realize they have access to other accounts
• Monitor admin users: Regularly review who has elevated access
• Compare QoS: Different QoS levels provide different priorities and limits
• Filter by account: Use
  Copy

  F3

  with
  Copy

  account:name

  to see all users in a research group
• Sort by usage limits: Identify power users with high CPU/node limits
• Default account matters: Billing and priority often depend on the default account
• Unlimited ≠ infinite: Still subject to partition and cluster limits

Fair-Share and Priority

User priority in the scheduler depends on:

1. Account Association Priority: Set per user-account pair
2. Fair-Share Score: Based on historical usage vs. allocation
3. QoS Priority: Different QoS levels have different priorities
4. Job Age: Older pending jobs get priority boost

View the Jobs view to see how priority affects actual job scheduling.

Common Issues

"Job submission failed: User limit exceeded"

• Check Max Jobs limit
• Wait for running jobs to complete
• Contact admin to request limit increase

"Job submission failed: CPU limit exceeded"

• Check Max CPUs limit
• Current jobs are using up allocation
• Reduce CPUs requested or wait for jobs to finish

"Cannot submit to account X"

• User not associated with account X
• Check Associated Accounts list
• Contact admin to request account access

"QoS not allowed"

• User trying to use QoS not in Allowed QoS list
• Specify a different QoS or use default
• Contact admin to request QoS access