How to Manage Slurm Accounts with 'sacctmgr' (with examples)

The sacctmgr command is a powerful tool for managing accounting data in systems using the Slurm workload manager. Slurm is widely used in high-performance computing environments for cluster management and job scheduling. sacctmgr allows administrators to handle various accounting tasks such as viewing current configurations, adding clusters or accounts, and displaying detailed information about users and associations. Below, we explore several use cases that demonstrate how sacctmgr can be effectively used to manage Slurm accounts.

Use case 1: Show Current Configuration

Code:

sacctmgr show configuration

Motivation:

Understanding the current configuration of a Slurm accounting system is crucial for system administrators. This command helps verify the setup and understand the environment’s operational parameters. Reviewing configurations can aid in troubleshooting, optimizing system performance, and ensuring correct resource allocations.

Explanation:

• sacctmgr: Invokes the Slurm account manager utility.
• show: A subcommand used to display information from the database.
• configuration: Specifies that the current configuration settings of the Slurm accounting system should be displayed.

Example Output:

ConfigurationName  Value
------------------  -----
ClusterName        my_cluster
AccountingStorage  ...
SchedulerType      ...

This output illustrates key configurations such as the cluster name and accounting storage parameters.

Use case 2: Add a Cluster to the Slurm Database

Code:

sacctmgr add cluster cluster_name

Motivation:

Adding a new cluster to the Slurm database is essential when expanding the computing infrastructure. This ensures that the new cluster is integrated into the overall system for management and resource scheduling. Properly adding a cluster helps in maintaining the seamless operation of diverse computing resources under a single management system.

Explanation:

• sacctmgr: Utilizes the Slurm account manager for database operations.
• add: Indicates an addition operation to the database.
• cluster: Specifies the target entity type to add.
• cluster_name: The name of the new cluster to be added, which should be unique within the database.

Example Output:

Adding cluster 'cluster_name'

This confirms that the specified cluster has been successfully added to the Slurm database.

Use case 3: Add an Account to the Slurm Database

Code:

sacctmgr add account account_name cluster=cluster_of_account

Motivation:

Creating accounts for users or projects within a cluster is necessary for resource allocation and tracking usage. By adding an account, administrators can assign resources, set limits, and monitor usage, leading to better resource management and accountability within the cluster.

Explanation:

• sacctmgr: The command interface used for managing accounting data.
• add: Specifies an addition operation to be performed.
• account: Denotes that an account is being added.
• account_name: The designated name for the account, typically signifying a user or project group.
• cluster=cluster_of_account: Associates the account with a specific cluster, ensuring that it is only valid within the context of this cluster.

Example Output:

Adding account 'account_name' to cluster 'cluster_of_account'

This output indicates that the account has been successfully created and linked to the specified cluster.

Use case 4: Show Details of User/Association/Cluster/Account Using a Specific Format

Code:

sacctmgr show user|association|cluster|account format="Account%10" format="GrpTRES%30"

Motivation:

Displaying detailed information about users, associations, clusters, or accounts in a customized format is crucial for monitoring and reporting. This flexibility allows administrators to focus on specific fields of interest and adjust the presentation to suit various administrative, troubleshooting, or auditing tasks.

Explanation:

• sacctmgr: The command tool for handling Slurm accounting.
• show: Indicates that information from the database should be displayed.
• user|association|cluster|account: Offers a choice between displaying information about users, associations, clusters, or accounts.
• format="Account%10": Sets the display format of the “Account” field, ensuring a column width of 10 characters.
• format="GrpTRES%30": Formats the “GrpTRES” field with a column width of 30 characters, allowing easy readability of resource allocations.

Example Output:

 Account     GrpTRES
----------  ------------------------------
 my_account  cpu=10,mem=50G
 ...

This output showcases how the account and corresponding group total resources are displayed, with specified formatting for clarity and readability.

Conclusion:

The sacctmgr command is indispensable for managing the diverse aspects of Slurm’s accounting system. It provides administrators with essential functionalities such as viewing configurations, adding clusters or accounts, and customizing data views. Each use case illustrates how sacctmgr can be utilized to maintain an efficient and orderly computing environment in high-performance computing contexts.