Job Count across the Reports

Job count qualification varies in different charts for the same time range depending on the status of the jobs.

The job count difference is based on the query used to generate the reports.

The time range query used for the job count based on the job status is as follows:
  • Running Jobs: the job count is based on the number of jobs that were running during the time window. The query to calculate the job count is:
    endTime >= FILTER_START_TIME AND startTime <= FILTER_END_TIME
    The above query is used in the following reports:
    • Cluster Usage
    • Node Hours
    • CPU Hours
    • GPU Hours
    • Memory Requested
  • Queued Jobs: the job count is based on the number of jobs that were pending during the time window. The query to calculate the job count is:
    startTime >= FILTER_START_TIME AND submitTime <= FILTER_END_TIME
    The above query is used in the following reports:
    • Time Spent Queuing
  • Completed Jobs: the job count is based on the number of jobs that ended during the time window. The query to calculate the job count is:
    endTime >= FILTER_START_TIME AND endTime <= FILTER_END_TIME
    The above query is used in the following reports:
    • Number of Jobs Completed
    • Cloud Cost
    • Unique Users
    • Expansion Ratio
    • Idle Jobs
    • Wallclock Requested vs. Used
    • Memory Requested vs. Used
    • Job Runtime