How zhongfucheng3y/austin works
Project type and purpose
• This is a full-stack, modular, event-driven message-push platform (“austin”) built on Spring Boot.
• It exposes a web UI and REST APIs for template/channel management, then internally assembles message tasks, publishes them to message queues, processes them in a handler pipeline, pushes to third-party channels, and provides real-time tracing and data analytics via Flink/Hive.
File-structure–based component breakdown
• austin-web: Spring Boot web layer (UI + REST controllers)
• austin-service-api + service-api-impl: service-layer interfaces and pipeline orchestration (pre-check → assemble → MQ → post-check)
• austin-common: shared domain, DTOs, enums, and pipeline engine abstractions
• austin-support: infrastructure adapters (Redis, RabbitMQ, Kafka, RocketMQ, EventBus), thread pools, DAO for MySQL
• austin-handler: consumption layer – receives tasks from MQs, runs business-process pipeline actions (deduplication, shield, send, etc.), calls channel SDKs/APIs
• austin-cron: scheduled task module (xxl-job integration) for batch/timed sends
• austin-stream: Flink job – consumes business/log topics from Kafka, writes metrics to Redis (real-time dashboards) and can sink to Hive
• austin-data-house: Hive write-side for long-term analytics; Flink job bootstrap
• Infra/config: docker-compose definitions (MySQL, Redis, Kafka, RabbitMQ, RocketMQ, Prometheus, Grafana, Graylog, Flink, Hive, Apollo/Nacos, metabase, xxljob)
Key README insights
• Core features: multi-channel messaging, template management, full-link tracing, cron jobs, de-dup, night shield, dynamic routing
• Tech stack: Java 8, Spring Boot 2.5.x, MySQL, Redis, Kafka/RabbitMQ/RocketMQ, Flink, Hive, xxl-job, Apollo/Nacos, Graylog, Prometheus/Grafana, Docker
Elements
• Clear, consistent labels for each module and external system
• Green clouds for message brokers (Kafka, RMQ, RocketMQ)
• Purple parallelograms for data pipelines (Flink)