?
HPC TaskMaster - Task Efficiency Monitoring System for the Supercomputer Center
This paper is devoted to the monitoring system HPC TaskMaster developed at the HSE University for the cHARISMa cluster. This system automatically evaluates the efficiency of performing tasks of HPC cluster users and identifies inefficient tasks, thereby significantly saving expensive machine time. In addition, users can view reports on completing their tasks, along with inferences about their work and interactive graphs. Particular attention in this paper is paid to determining the effectiveness of the task - the system allows the administrator to personally configure the criteria for evaluating the effectiveness of the task without the need for changes in the source code. The system is developed using open-source software and is publicly available for use on other clusters.