A Realtime Monitoring Method for Cluster System Running State Based on Network
Wanqing Chi, Wei Hu Zhou · Journal of Physics Conference Series · 2019
High Performance Computing (HPC) system running state monitoring is an important task. Small-scale cluster systems exist separately as one whole system or exist as a part of a large-scale HPC system. Such cluster systems are often the core of the entire system. Therefore, for such systems, there should be more timely and accurate monitoring of running state. However, various HPC system monitoring softwares currently focus on the aggregation of overall running and resource state, and cannot provide realtime monitoring, detailed information, and process control. Based on the existing single-machine real-time monitoring tools, this paper adds network-based information exchange protocol, multi-threaded parallel information acquisition service, extends information representation and display output style, adds remote process control, then carries out related tests, which implements a more efficient real-time running state monitoring tool for small-scale cluster systems.