Telemetry datasets for AI-driven self-healing of cloud-native microservices (Online Boutique, Sock Shop, Train Ticket)

Metrics, logs, distributed traces and Kubernetes events recorded from three open-source microservice benchmarks (Online Boutique, Sock Shop and Train Ticket) while faults were injected with Chaos Mesh. Each dataset comes with ground-truth injection logs. Two campaigns were run per benchmark: an anomaly campaign with six fault types (cpu, mem, disk, delay, loss, socket) injected into five services, and a failure campaign with 100% packet loss (SLA failures). The data were collected for the MSc dissertation "An AI-Driven Autonomous Self-Healing System for Cloud-Native Microservice Applications" (Master's in Software Engineering, Faculty of Engineering of the University of Porto, FEUP) and the journal paper derived from it. See README.md for the file structure, formats and known limitations.