SRE & observability
Reliability Intelligence Platform
A flagship SRE and observability project built around a FastAPI Orders API with production-style instrumentation. A local Docker Compose stack runs the API, Prometheus, and Grafana, and the service evaluates its own rolling SLI/SLO compliance and error budget, then detects and manages incidents against those objectives.
Stack
- Python
- FastAPI
- Docker
- Docker Compose
- Prometheus
- Grafana
- pytest
219 tests passing (verified locally).
System map
verified stack
docker compose · local network
Orders API:8000
- health · readiness
- structured logging
- error handling
Prometheus:9090
- scrapes the API
- time-series storage
Grafana:3000
- provisioned datasource
- service overview
Metrics instrumented in-apprequest raterequest latencyerror rate · 5xx
Build log — verified
- 01FastAPI Orders API with separate health and readiness endpoints, structured logging, error handling, Docker packaging, and automated tests.
- 02Prometheus application metrics exposed through a /metrics endpoint with low-cardinality labels.
- 03Docker Compose stack with Prometheus scraping the API and a provisioned Grafana service-overview dashboard.
- 04Guarded latency and HTTP 500 failure injection, disabled by default and refused in production.
- 05Rolling-window SLI/SLO evaluation for availability and latency, exposed as JSON at /api/reliability.
- 06Error budget derived from each SLO and exported as Prometheus gauges and Grafana panels.
- 07Automatic incident detection that opens, updates, and resolves incident records from the same reliability window, exposed through a read-only /api/incidents API with an acknowledge action.
- 08Reliability panels report INSUFFICIENT DATA rather than a false green when the observation window is empty.