High-Performance Computing (HPC) systems require robust and efficient monitoring solutions to ensure reliability, resource optimization, and rapid anomaly detection. This paper presents the design and implementation of a lightweight, modular monitoring framework for the Mahameru HPC infrastructure operated by the National Research and Innovation Agency (BRIN) of Indonesia. The proposed system integrates shell-based monitoring probes with the Elastic Stack (ELK) to provide real-time insights into CPU, memory, disk status, user activities, and system vulnerabilities. It features a dual-threshold automatic kickout mechanism that enforces resource usage limits by terminating excessive user processes and notifying administrators through an automated email pipeline. Evaluation results show the system achieves 99.41% measurement accuracy with negligible performance overhead (2.8% of the monitoring interval), reduces CPU load by over 99% in overload scenarios, and delivers notifications with a 0.65-second average delay. This work demonstrates a practical approach to proactive resource management in national-scale HPC environments. © 2025 IEEE.