HDFS-8545. Refactor FS#getUsed() to use ContentSummary and add an API to fetch the total file length from a specific path (Contributed by J.Andreina)

(cherry picked from commit 7d2d16f4ee)

hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/fs/FileSystem.java

@@ -2067,12 +2067,13 @@ public abstract class FileSystem extends Configured implements Closeable {
 
   /** Return the total size of all files in the filesystem. */
   public long getUsed() throws IOException {
-    long used = 0;
-    RemoteIterator<LocatedFileStatus> files = listFiles(new Path("/"), true);
-    while (files.hasNext()) {
-      used += files.next().getLen();
+    Path path = new Path("/");
+    return getUsed(path);
   }
-    return used;
+
+  /** Return the total size of all files from a specified path. */
+  public long getUsed(Path path) throws IOException {
+    return getContentSummary(path).getLength();
   }
 
   /**

hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/fs/FilterFileSystem.java

@@ -390,6 +390,12 @@ public class FilterFileSystem extends FileSystem {
     return fs.getUsed();
   }
 
+  /** Return the total size of all files from a specified path.*/
+  @Override
+  public long getUsed(Path path) throws IOException {
+    return fs.getUsed(path);
+  }
+
   @Override
   public long getDefaultBlockSize() {
     return fs.getDefaultBlockSize();

hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/fs/HarFileSystem.java

@@ -1237,6 +1237,12 @@ public class HarFileSystem extends FileSystem {
     return fs.getUsed();
   }
 
+  /** Return the total size of all files from a specified path.*/
+  @Override
+  public long getUsed(Path path) throws IOException {
+    return fs.getUsed(path);
+  }
+
   @SuppressWarnings("deprecation")
   @Override
   public long getDefaultBlockSize() {

hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt

@@ -766,6 +766,9 @@ Release 2.8.0 - UNRELEASED
     HDFS-9295. Add a thorough test of the full KMS code path. 
     (Daniel Templeton via zhz)
 
+    HDFS-8545. Refactor FS#getUsed() to use ContentSummary and add an API to fetch
+    the total file length from a specific path (J.Andreina via vinayakumarb)
+
   OPTIMIZATIONS
 
     HDFS-8026. Trace FSOutputSummer#writeChecksumChunks rather than

hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestDistributedFileSystem.java

@@ -1151,4 +1151,32 @@ public class TestDistributedFileSystem {
       cluster.shutdown();
     }
   }
+
+  @Test(timeout = 30000)
+  public void testTotalDfsUsed() throws Exception {
+    Configuration conf = new HdfsConfiguration();
+    MiniDFSCluster cluster = null;
+    try {
+      cluster = new MiniDFSCluster.Builder(conf).numDataNodes(1).build();
+      FileSystem fs = cluster.getFileSystem();
+      // create file under root
+      FSDataOutputStream File1 = fs.create(new Path("/File1"));
+      File1.write("hi".getBytes());
+      File1.close();
+      // create file under sub-folder
+      FSDataOutputStream File2 = fs.create(new Path("/Folder1/File2"));
+      File2.write("hi".getBytes());
+      File2.close();
+      // getUsed(Path) should return total len of all the files from a path
+      assertEquals(2, fs.getUsed(new Path("/Folder1")));
+      //getUsed() should return total length of all files in filesystem
+      assertEquals(4, fs.getUsed());
+    } finally {
+      if (cluster != null) {
+        cluster.shutdown();
+        cluster = null;
+      }
+    }
+  }
+
 }