Hadoop回收站文件夹配置：如何设置/修改自定义路径？

Great question! Let's break down how to change Hadoop's default trash directory from the standard /user/<computer-name>/.Trash (on HDFS) to a custom path—either on HDFS or the local filesystem (LFS). Here's everything you need to know:

1. Modify to a Custom HDFS Path #

Hadoop manages trash settings via the core-site.xml configuration file. To set a custom HDFS path:

• Open your cluster's core-site.xml file (typically located in $HADOOP_HOME/etc/hadoop/).
• Add or update the following configuration properties:

  <property>
    <name>fs.trash.root</name>
    <value>/custom/hadoop/trash/${user.name}</value>
    <description>Custom HDFS path for user-specific trash directories. ${user.name} ensures each user gets their own isolated subfolder.</description>
  </property>
  <!-- Optional: Adjust trash retention time (in minutes) if needed -->
  <property>
    <name>fs.trash.interval</name>
    <value>1440</value> <!-- 24 hours; default is 0 which disables trash -->
  </property>
  

• Save the file and restart Hadoop's NameNode and DataNode services to apply changes:

  # Stop HDFS services
  $HADOOP_HOME/sbin/stop-dfs.sh
  # Restart HDFS services
  $HADOOP_HOME/sbin/start-dfs.sh
  

• Critical Permissions Step: Create the parent HDFS directory and set proper permissions so all users can access their trash:

  hdfs dfs -mkdir -p /custom/hadoop/trash
  hdfs dfs -chmod 777 /custom/hadoop/trash
  

2. Modify to a Local Filesystem (LFS) Path #

If you want the trash to live on the local filesystem (e.g., for Hadoop running in local mode), specify a local path with the file:// prefix:

• Edit core-site.xml and update the fs.trash.root property:

  <property>
    <name>fs.trash.root</name>
    <value>file:///opt/hadoop/local-trash/${user.name}</value>
    <description>Local filesystem path for user-specific trash. The triple slash denotes an absolute local path.</description>
  </property>
  

• Save the file and restart any running Hadoop services (like the local resource manager if using YARN).
• Local Permissions Check: Create the parent local directory and set permissions for all users:

  sudo mkdir -p /opt/hadoop/local-trash
  sudo chmod 777 /opt/hadoop/local-trash
  

Quick Notes #

• For single-user overrides (not cluster-wide), a user can create a ~/.hadoop/core-site.xml file with the above properties—this will take precedence over the cluster's global config for that user.
• Verify the change works by deleting a file with hdfs dfs -rm <hdfs-file-path> (for HDFS) or hadoop fs -rm <local-file-path> (for LFS), then checking your custom trash directory to confirm the file was moved there.

内容的提问来源于stack exchange，提问作者Anand