Cloudera QuickStart VM 5.12 Hive元存储故障：Spark-Shell启动告警求助

Hey there! Let's unpack what's going on here and answer your core question clearly.

先聊聊那个告警的原因 #

The WARN metastore.ObjectStore: Version information not found in metastore... warning you're seeing happens because when Hive/Spark first connects to the MySQL metastore, it can't find an existing schema version record. Since hive.metastore.schema.verification is disabled (default in many quick setups), it just logs this warning and creates the version entry on the fly. It's not a critical error, but totally confusing when you expect a "Quick Start" environment to work out of the box—totally get that frustration!

核心问题：用MySQL做Hive元存储时，需要启动Hive metastore服务吗？ #

Short answer: Yes, you should (and almost always need to) run the Hive metastore service when using MySQL as your Hive metastore in Cloudera environments. Here's why:

Cloudera's setup relies on the Hive metastore service (a Thrift server) as the intermediary between clients (like Spark) and the MySQL database. There are two ways to access Hive metadata:

• Remote Metastore Mode (Recommended)：This is the standard setup for Cloudera. The Hive metastore service runs as a separate daemon, handles connections to MySQL, caches metadata, and enforces access controls. Spark connects to this Thrift service (via hive.metastore.uris configuration) instead of talking directly to MySQL. This avoids the schema warning you're seeing and ensures compatibility with all Cloudera components.
• Direct JDBC Mode (Not Recommended)：You could technically configure Spark to connect directly to MySQL via JDBC, but this skips the metastore service entirely. This lacks caching, breaks some Hive features, and isn't supported in most Cloudera deployments—definitely not ideal for even a Quick Start environment.

怎么解决你的问题 #

1. Check if the Hive metastore service is running：Open Cloudera Manager (the web UI included in Quick Start), navigate to the Hive service, and make sure the "Hive Metastore Server" instance is started. If it's stopped, start it up.
2. Verify your Hive configuration：Ensure hive.metastore.uris is set to point to the metastore service (e.g., thrift://localhost:9083). This tells Spark to use the metastore service instead of directly connecting to MySQL.
3. Initialize the Hive metastore schema (if needed)：If this is your first time using MySQL for the metastore, run the schema initialization command:

   schematool -dbType mysql -initSchema
   

   This creates all the necessary tables in MySQL, including the version record that's missing (which will make the warning go away).

Once you've got the metastore service running and the schema initialized, that warning should disappear, and Spark should interact with the Hive metastore properly.

内容的提问来源于stack exchange，提问作者Ged