Select a Report

Select a report from the list to generate the report chart and to view its metrics.

The Metric changes based on the report that is selected. For example, the following Metric is displayed for the Cluster Usage Report:

Number of Running Jobs
CPU hours
GPU hours
Memory in GB hours
Node hours

Select a report from the drop-down list to generate a report.

The following table provides the list of report that is available:


Report Name	Description	Metric Details
Cluster Usage	Shows overall cluster usage for resource utilization as well as the number of jobs that were running during the selected time window. If the total resources available is listed in the database then the chart will show the resources used as a percentage of resources available and there will be a data series showing unused resources. Otherwise the resources used by each data series will be shown as a percentage of the total resources used. When there are more than 20 data series in the chosen grouping the chart will show the top 20 data series and group the rest into one data series called "other". In the table, all data series will be listed.	Number of Running Jobs CPU Hours GPU Hours Memory in GB Hours Node Hours
Cloud Cost	Shows the cloud cost as calculated by the instance cost and proportion of the cloud instance used by the job as a proportion of the cores used. This value is also known as the burn rate. The jobs are accounted for at the end time. The actual amount billed by the cloud provider may vary from the estimated cost shown here.	Number of Completed Jobs Total Cost (USD)
Number of Jobs Completed Over Time	This report shows the number of jobs completed during the time window selected. This report includes jobs that failed for any reason as well as jobs that succeeded.	Number of Completed Jobs
Number of Jobs Completed	This report shows the number of jobs completed during the time window selected. This report includes jobs that failed for any reason as well as jobs that succeeded.	Number of Completed Jobs
Node Hours	This report shows Node hours for jobs that have completed: *(Frac's of Nodes used) wallclockTimeUsed. (Frac's of Nodes used) = (slots used in the Nodes) /(slots available in the Nodes)** Incase job ran with exclusive flag, 'Frac's of Nodes used' will be 1. All jobs that are running during the time window are included in the report provided they have now completed.	Node Hours Number of Running Jobs
CPU Hours	Shows CPU hours: *slots wallclockTimeUsed**. All jobs that are running during the time window are included in the report provided they have now completed.	CPU Hours Number of Running Jobs
GPU Hours	Shows GPU hours: *gpusRequested wallclockTimeUsed**. All jobs that are running during the time window are included in the report provided they have now completed.	GPU Hours Number of Running Jobs
Memory Requested	Shows total memoryRequested * wallclockTimeUsed. All jobs that are running during the time window are included in the report provided they have now completed.	Memory in GB Hours Number of Running Jobs
Unique Users	Shows the number of unique users for jobs that completed during the time window selected.	Unique User Count Number of Completed Jobs
Time Spent Queuing	Shows the time jobs spent pending between job submission and job start: startTime - submitTime All jobs that were pending during the time window are included in the report provided they have now completed.	Time Spent Queuing (Hr) Number of Queueing Jobs
Expansion Ratio	Shows the expansion ratio: (endTime - submitTime) / wallclockTimeUsed Expansion Ratio is the ratio of the time spent pending + running against the time spent running. Shows the number of jobs with Expansion Ratio in the given range for jobs that completed during the time window selected.	Expansion ratio Number of Completed Jobs
Idle Jobs	Shows jobs where the the cpuTimeUsed is very low vs. the wallclockTimeUse for jobs that completed during the time window selected. Reasons can be: The job may have hung in deadlock The job may have been waiting for a resource such as a license One or more processes may have crashed leaving the job still running The job may be interactive and left running, forgotten by the user The job may have requested many slots, but is only using one core	Number of Completed Jobs Idle Jobs (CPU utilization <2%) Idle Jobs (CPU utilization <10%)
Wallclock Requested vs. Used	Compares the wallclock requested as a percentage of wallclock used. Shows the number of jobs with a percentage of wallclockTimeUsed vs wallClockTimeRequested in the given range.	Number of Completed Jobs % Wallclock Used
Memory Requested vs. Used	Compares the memory requested as a percentage of RSS memory used. Shows the number of jobs with a percentage of physicalMemoryUsed vs memoryRequested in the given range.	Number of Completed Jobs % Memory Used
Job Runtime	Shows wallclockTimeUsed grouped into runtime buckets for jobs that completed during the time window selected.	Number of Completed Jobs Job Runtime
Job Summary Report	Provides a table of primary metrics per job for jobs that completed during the time window selected. The table will only show the first 1000 results. Please set filters to narrow down the results.	Cluster Job Id Array Index Owner Project Job Name CPU Requested Memory Requested GB GPU Requested GPU Hours Expansion Ratio CPU Used(Hr) Memory Used(GB) Start Time End Time Run Time(Hr) Submit Time Queue Time(Hr)