yarn hadoop Hadoop airflow mysql dag ops troubleshooting operator ranger ambari exception hdfs Ambari flink others kafka atlas spark hive flume ClickHouse olap clickhouse tableEngine metaData clickhouseFunctions deepagent dolphinScheduler Flink kerberos Greenplum greenplum mapreduce sqoop datanode hbase HBase Hive hiveFunctions java other idea JVM hudi data-lake lakehouse big-data cow mor query compaction index bloom bucket timeline instant filegroup fileslice write upsert insert delete spark-sql flink-sql checkpoint rocksdb deltastreamer hoodie-streamer concurrency tuning lock performance internals clean changelog clustering catalog Kafka hive-sync metastore avro langchain langgraph langsmith agent openai qwen mlx message llm 大模型 prompt output-parser structured-output pydantic multimodal mlx-vlm gemma gradio mlx-whisper speech-to-text cli chatbot memory sqlite workflow cdc streaming studio api sdk sse agentstate tool context configurable postgresql asyncio evaluator state reducer mcp human-in-the-loop interrupt command json-rpc fastmcp python spring-ai security jwt oauth cloudstack keycloak storm hexo centos linux Zookeeper fastdfs azkaban Iterm2 yum rpm markdown Phoenix phoenix prometheus monitoring grafana 爬虫 rag embeddings one-hot cosine-similarity 向量 embedding semantic-search pandas vector-database faiss chroma milvus vector-search qwen3 sentence-transformers apple-silicon retriever redis cache nosql docker rdb aof transaction pipeline pubsub datatype sentinel high-availability replica replication io epoll cluster springboot bigkey canal bitmap hyperloglog geo bloom-filter availability redisson redlock lua source-code data-structure Scala Maven Spark sparkCore sparkStreaming spark streaming spark sql sparkSQL sparkMl spark ml parquet