Hive2.3.3+Hadoop3.1.0创建表失败：元数据缺失及SLF4J绑定冲突

Hey there, let's tackle the issues you're hitting one by one. I've dealt with these exact problems before, so here's how to fix each error step by step:

1. Fix the readlink: illegal option -- f Error #

This pops up because macOS uses BSD-style readlink which doesn't support the -f flag, but Hive's startup scripts expect the GNU version of readlink. Here's the fix:

• Install GNU coreutils via Homebrew (if you don't have Homebrew, set it up first):

  brew install coreutils
  

• Locate your Hive installation's hive script (usually in $HIVE_HOME/bin), open it in a text editor, and replace every instance of readlink -f with greadlink -f (GNU readlink is named greadlink on macOS).

2. Resolve SLF4J Multiple Bindings Warning #

The SLF4J warning happens because both Hive and Hadoop include their own SLF4J implementation JARs, causing a conflict. To fix this:

• Navigate to your Hadoop installation's share/hadoop/common/lib directory.
• Find the JAR file named something like slf4j-log4j12-*.jar (the version number might vary) and either rename it (e.g., to slf4j-log4j12-*.jar.bak) or move it to a backup folder. This removes the conflicting binding from Hadoop's classpath, letting Hive's preferred implementation take over.

3. Fix MetaException: Version information not found in metastore #

This is the critical issue stopping you from creating tables—it means your Hive metastore hasn't been initialized yet. Here's how to fix it:

• First, make sure your Hadoop cluster is up and running. Start HDFS and YARN with these commands:

  start-dfs.sh
  start-yarn.sh
  

• Initialize the metastore using Hive's schematool. Since you're using the default Derby database, run:

  schematool -dbType derby -initSchema
  

  (If you're using a different database like MySQL for the metastore, replace derby with your database type and ensure you've configured the JDBC settings in hive-site.xml first.)
• After running the initialization command, restart the Hive shell and try creating your table again—it should work now.

内容的提问来源于stack exchange，提问作者claudio gugliotta