Centralised Log Monitoring & Alerting System for Microservices

Problem Statement & Abstract

When an application runs as many services, finding the cause of an error means searching through logs scattered across machines. This project builds the observability stack that operations teams rely on: every log line and metric flows into one place where it can be searched, visualised and turned into alerts.

A sample shop made of Python and Node.js services, with Redis as a cache, runs in Docker Compose. Each service emits structured JSON logs with a request ID, so one user request can be followed across services. Promtail ships the logs to Grafana Loki, while Prometheus scrapes metrics such as request count, latency percentiles and cache hit rate.

Grafana dashboards combine logs and metrics side by side. Alert rules notify the team by email or Telegram when error rates rise, latency passes a threshold or a service stops responding, and every alert links to the relevant logs. A fault-injection script deliberately breaks services during the demo so you can show detection, alerting and diagnosis end to end.