What problem does it solve?
It helps you reliably orchestrate StarRocks ingestion workflows from Airflow, including async Broker Load polling and Routine Load lifecycle management, so ETL pipelines don’t get stuck or load with stale metadata.
Core Features & Use Cases
- Broker Load orchestration (S3 batch ingestion): Create partitions, trigger Broker Load jobs with a deterministic label, and poll
SHOW LOAD ... until FINISHED.
- Stream Load micro-batch pattern: Send NDJSON payloads via HTTP with a computed label and validate StarRocks response status.
- Routine Load lifecycle control: Pause for schema changes, apply DDL, and resume while handling job state transitions safely.
- Post-load optimization: Run
ANALYZE TABLE for partitions to refresh statistics and improve query planning.
- Airflow integration details: Use
MySqlHook for MySQL-compatible DDL/DML and Http/requests for Stream Load; pass labels via XCom and template partition-aware DAG parameters.
Quick Start
Ask the agent to generate an Airflow DAG that performs partition-aware Broker Load into a StarRocks table, polls until completion, then runs ANALYZE TABLE ... PARTITION (...) and basic row-count validation for the same date.