Deployment and operation of AUDITOR use cases

Research conducted by the University of Freiburg.

 

We worked on the use cases listed below.

Priority plugin use case

In this use case, four local HEP research groups (A to D) share resources in the NEMO project. Each group is managed by its own COBalD/TARDIS instance and has a dedicated SLURM partition (job queue). By sharing virtual machines across the HEP groups, resources are used efficiently. Additionally, the Collector-AUDITOR-Plugin pipeline is utilized to adjust group priorities on ATLAS-BFG.

Use case results:

  • Resources of the four local HEP groups have been provided
  • Priority is adjusted according to the provided resources

APEL Accounting with AUDITOR use case

During this use case:

  • dedicated AUDITOR pipeline was operated
  • accounting data from SLURM was collected
  • data was stored as records in AUDITOR db
  • APEL plugin retrieved records from AUDITOR
  • APEL job summary records were created
  • summary to APEL test instance was sent.

Data collection for Utilization Report Tool

During this use case:

  • dedicated AUDITOR pipeline was operated
  • accounting data from SLURM was collected
  • data was stored as records in AUDITOR db
  • data was provided via pyauditor in jupyter nb

Support

During this period, we have supported both users and developers of AUDITOR, and our main focus was on the support efforts. We also provided support for COBalD/TARDIS components delivered by Freiburg. Besides, regular accounting meetings were held.

Future work

We plan to work on the following aspects in the future:

  • AUDITOR DB stress tests. We will analyze the results of AUDITOR DB stress tests and adjust components according to these results.
  • AUDITOR components. We will work on further development and adjustment of AUDITOR components to requirements
  • APEL Accounting with AUDITOR use case. For this use case, successful accounting in EGI production instance is planned.
  • Data collection for utilization report tool. At the moment, the data has been mostly collected only in this use case. In the future, the data is going to be analysed.
  • Support. We plan to improve documentation and continue improving the current support process.
Cookie Consent mit Real Cookie Banner