Speaker
Description
The ATLAS distributed computing system has allowed the experiment to successfully meet the challenges of LHC Run 2. In order for distributed computing to operate smoothly and efficiently, several support teams are organized in the ATLAS experiment. The ADCoS (ATLAS Distributed Computing Operation Shifts) is a dedicated group of shifters who follow and report failing jobs, failing data transfers between sites, degradation of ATLAS central computing services, and more. The DAST (Distributed Analysis Support Team) provides user support to resolve issues related to running distributed analysis on the grid. The CRC (Computing Run Coordinator) maintains a global view of the day-to-day operations.
In this presentation, the status and operational experience of the support system for ATLAS distributed computing in LHC Run 2 will be reported. This report also includes operations experience from the grid site point of view, and an analysis of the errors that create the biggest waste of wallclock time. The report of operations experience will focus on some of the more time-consuming tasks for shifters and grid sites, and on the introduction of new technologies, such as machine learning, to ease the work.
