hdfs-tree is a utility tool to display hdfs directory in tree structure, meanwhile you can sort directory by size/file count/mtime. This is useful when you want to check the disk usage of hdfs, just like what we do in bash with du -h --max-depth=1 / | sort -hr | head to check the linux file system disk usage.
-
cd ${Project_HOME} && mvn clean -DskipTests package
-
cd ${Project_HOME}/target && java -jar hdfs-tree_2.10-1.0.jar xxx
Possible parameters you can pass with hdfs-tree are as follows:
Usage: hdfs-tree [OPTION]
Options:
-d, --depth <arg> max traverse depth for directory, default 5
-l, --limit <arg> top N to display by sort, default 5
-r, --reverse reverse the sort or not
-s, --sort <arg> sort options, must be size | mtime | count(file count)
-t, --type <arg> file type to compare, d | f
--help Show help messageHowever, if you want to traverse the root directory, you should use hdfs://localhost/ instead of hdfs://localhost。 As for the reason, please check org.apache.hadoop.hdfs.DistributedFileSystem.getPathName and everything will be clear.
In hadoop cluster that supports namenode ha, you should pass the hadoop configuration directory in the class path, i.e
java -cp ${HADOOP_CONF_DIR}:hdfs-tree_2.10-1.0.jar com.jacoffee.HDFSTree $@Also, you can configure two environment variables: ${HDFS_CONF_DIR}、${HDFS_TREE_HOME} and use the script hdfs-tree in ${HDFS_TREE_HOME}/bin directly.
java -jar hdfs-tree_2.10-1.0.jar -r -l 3 -d 2 -s size hdfs://localhost/hdfs://localhost/
├── [ 2GB] user
├── [ 2GB] allen
├── [ 526MB] hive
├── [ 1KB] platform
├── [ 709MB] tmp
├── [ 709MB] hadoop-yarn
├── [ 0B] hive
├── [ 99MB] spark
├── [ 99MB] eventLogjava -jar hdfs-tree_2.10-1.0.jar -r -l 3 -d 2 -s count hdfs://localhost/hdfs://localhost/
├── [ 692] spark
├── [ 692] eventLog
├── [ 392] tmp
├── [ 392] hadoop-yarn
├── [ 0] hive
├── [ 301] user
├── [ 201] hive
├── [ 99] allen
├── [ 1] platformjava -jar hdfs-tree_2.10-1.0.jar -r -l 3 -d 2 -s mtime hdfs://localhost/hdfs://localhost/
├── [2018-02-05 09:44:07] user
├── [2018-04-02 22:50:19] allen
├── [2018-02-05 09:44:07] platform
├── [2017-08-06 11:18:50] hive
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent