How to use the command 'sprio' (with examples)

The ‘sprio’ command is a Slurm utility that allows users to view the factors determining the scheduling priority of jobs in a Slurm job scheduler. It provides valuable information for understanding how the scheduler assigns priorities to jobs, which can help users optimize their job submission strategy.

Use case 1: View the factors determining the scheduling priority of all jobs

Code:

sprio

Motivation: This use case is helpful when users want to get a general overview of the scheduling priorities of all jobs in the Slurm job scheduler. By viewing the factors influencing the priorities, users can determine if there are any bottlenecks or imbalances in the scheduling that might affect job execution.

Explanation: The command ‘sprio’ without any arguments will display the factors determining the scheduling priority of all jobs. It will provide information such as job ID, partition, priority, reason, and any related dependencies.

Example output:

JOBID PARTITION PRIORITY REASON           DEPENDENCY
1     debug     687      FairShare|
                           Priority|
                           Usage|
                           Age
2     debug     692      FairShare|
                           Priority|
                           Usage|
                           Age
3     debug     660      FairShare|
                           Priority|
                           Usage|
                           Age
...

Use case 2: View the factors determining the scheduling priority of specified jobs

Code:

sprio --jobs=job_id_1,job_id_2,...

Motivation: In some cases, users might want to specifically check the factors influencing the scheduling priority of certain jobs. This use case allows users to view the priority details of specified jobs, helping them understand why these jobs are scheduled in a particular order.

Explanation: By using the ‘–jobs’ option followed by a comma-separated list of job IDs, users can view the factors determining the scheduling priority of specific jobs. The command will display information for each specified job, such as partition, job priority, reason, and any related dependencies.

Example output:

JOBID PARTITION PRIORITY REASON           DEPENDENCY
1     debug     692      FairShare|
                           Priority|
                           Usage|
                           Age
2     debug     660      FairShare|
                           Priority|
                           Usage|
                           Age

Use case 3: Output additional information

Code:

sprio --long

Motivation: When users need more detailed information about the scheduling priorities, including the individual factor weights, this use case is useful. The additional information can provide insight into how the scheduler calculates the priority value for each job.

Explanation: The ‘–long’ option can be used to output additional information about the factors determining the scheduling priority of all jobs. This information includes the weight assigned to each factor, which affects the calculation of the priority. The command will display the factors, their weights, and the job IDs along with the other priority information.

Example output:

JOBID        WEIGHT   FACTOR          VALUE
1            1,1,1,1  FairShare       0,
                                       Usage        0,
                                       Priority     0,
                                       Age          0
2            1,1,1,1  FairShare       0,
                                       Usage        0,
                                       Priority     0,
                                       Age          0
3            1,1,1,1  FairShare       0,
                                       Usage        0,
                                       Priority     0,
                                       Age          0
...

Use case 4: View information for the jobs of specified users

Code:

sprio --user=user_name_1,user_name_2,...

Motivation: When users want to check the scheduling priority details for jobs owned by specific users, this use case is helpful. It allows users to focus on the jobs of interest and understand the influencing factors for those jobs only.

Explanation: By using the ‘–user’ option followed by a comma-separated list of user names, users can view the factors determining the scheduling priority for jobs owned by those users. The command will display information such as partition, priority, reason, and any related dependencies for each specified user’s jobs.

Example output:

JOBID PARTITION PRIORITY REASON           DEPENDENCY
1     debug     692      FairShare|
                           Priority|
                           Usage|
                           Age
3     debug     660      FairShare|
                           Priority|
                           Usage|
                           Age
...

Use case 5: Print the weights for each factor determining job scheduling priority

Code:

sprio --weights

Motivation: In some cases, users might be interested in knowing the specific weights assigned to each factor that contributes to the calculation of the scheduling priorities. By printing the weights, users can understand the influence of each factor and how their values contribute to the overall priority.

Explanation: The ‘–weights’ option can be used to print the weights assigned to each factor used in the calculation of job scheduling priority. The command will display the factors and their respective weights.

Example output:

FACTOR      WEIGHT
FairShare   0.25
Usage       0.75
Priority    1.00
Age         0.01

Conclusion:

The ‘sprio’ command is a powerful tool for viewing the factors determining the scheduling priority of jobs in a Slurm job scheduler. By utilizing different use cases, users can gain a deeper understanding of the job priority calculation and use that knowledge to optimize their job submission strategy.