Kakao Makers Builds Process-Level Monitoring Pipeline for Next.js SSR Servers

The Kakao Makers frontend team found that existing OS-level monitoring based solely on TSCoke could not detect heap memory limits (2–4GB) or event loop delays within Node.js processes, necessitating a monitoring system capable of verifying process status. They configured a pipeline that collects metrics with pm2-metrics, transmits them via Grafana Alloy, and stores them in TSCoke. Initially, they considered a distributed approach installing Alloy on each server, but to resolve the management overhead of 20+ servers and resource contention issues on SSR servers, they adopted a centralized Active-Standby HA configuration. Additionally, they ensured stability by managing the monitoring environment as Infrastructure as Code (IaC) using Docker Compose.