Ahmed Baha Eddine Alimi - B23-SD-01 [Lab07 + Bonus Task]

[3llimi]

Lab 07 — Observability & Logging with Loki Stack

What was done:

• Deployed centralized logging stack (Loki 3.0 + Promtail 3.0 + Grafana 12.3.1) via Docker Compose with shared logging network
• Configured Loki with TSDB storage backend (schema v13), 7-day retention, and compactor for automatic log cleanup
• Configured Promtail with Docker service discovery via socket — only scrapes containers with logging=promtail label, extracts app and container labels via relabeling
• Updated Python app to output structured JSON logs using a custom JSONFormatter — each request logs method, path, status_code, client_ip, duration_seconds as top-level JSON fields
• Integrated both apps (devops-python, devops-go) into the logging stack with Docker labels for Promtail filtering
• Built Grafana dashboard "DevOps Apps - Log Overview" with 4 panels: Logs Table, Request Rate (time series), Error Logs, and Log Level Distribution (pie chart)
• Configured production hardening: resource limits on all services, health checks for Loki and Grafana, Grafana anonymous access disabled, secrets in .env excluded from git
• Verified idempotency — second run produces changed=0

Bonus:

• Created roles/monitoring Ansible role with block/rescue/always structure, Jinja2-templated configs, and role dependency on docker via meta/main.yml
• All versions, ports, retention period, and resource limits parameterised in defaults/main.yml — fully configurable without touching role code
• Confirmed idempotency with community.docker.docker_compose_v2 — changed=0 on second run