diff --git a/hadoop-hdfs-project/hadoop-hdfs-rbf/src/main/java/org/apache/hadoop/hdfs/server/federation/router/RouterClientProtocol.java b/hadoop-hdfs-project/hadoop-hdfs-rbf/src/main/java/org/apache/hadoop/hdfs/server/federation/router/RouterClientProtocol.java
index baee979dfe7..ec61258f527 100644
--- a/hadoop-hdfs-project/hadoop-hdfs-rbf/src/main/java/org/apache/hadoop/hdfs/server/federation/router/RouterClientProtocol.java
+++ b/hadoop-hdfs-project/hadoop-hdfs-rbf/src/main/java/org/apache/hadoop/hdfs/server/federation/router/RouterClientProtocol.java
@@ -103,6 +103,7 @@ import java.io.IOException;
 import java.net.ConnectException;
 import java.util.ArrayList;
 import java.util.Collection;
+import java.util.Collections;
 import java.util.EnumSet;
 import java.util.HashMap;
 import java.util.Iterator;
@@ -771,13 +772,14 @@ public class RouterClientProtocol implements ClientProtocol {
 
     List<RemoteResult<RemoteLocation, DirectoryListing>> listings =
         getListingInt(src, startAfter, needLocation);
-    Map<String, HdfsFileStatus> nnListing = new TreeMap<>();
+    TreeMap<String, HdfsFileStatus> nnListing = new TreeMap<>();
     int totalRemainingEntries = 0;
     int remainingEntries = 0;
     boolean namenodeListingExists = false;
+    // Check the subcluster listing with the smallest name to make sure
+    // no file is skipped across subclusters
+    String lastName = null;
     if (listings != null) {
-      // Check the subcluster listing with the smallest name
-      String lastName = null;
       for (RemoteResult<RemoteLocation, DirectoryListing> result : listings) {
         if (result.hasException()) {
           IOException ioe = result.getException();
@@ -824,6 +826,10 @@ public class RouterClientProtocol implements ClientProtocol {
 
     // Add mount points at this level in the tree
     final List<String> children = subclusterResolver.getMountPoints(src);
+    // Sort the list as the entries from subcluster are also sorted
+    if (children != null) {
+      Collections.sort(children);
+    }
     if (children != null) {
       // Get the dates for each mount point
       Map<String, Long> dates = getMountPointDates(src);
@@ -838,9 +844,27 @@ public class RouterClientProtocol implements ClientProtocol {
         HdfsFileStatus dirStatus =
             getMountPointStatus(childPath.toString(), 0, date);
 
-        // This may overwrite existing listing entries with the mount point
-        // TODO don't add if already there?
-        nnListing.put(child, dirStatus);
+        // if there is no subcluster path, always add mount point
+        if (lastName == null) {
+          nnListing.put(child, dirStatus);
+        } else {
+          if (shouldAddMountPoint(child,
+                lastName, startAfter, remainingEntries)) {
+            // This may overwrite existing listing entries with the mount point
+            // TODO don't add if already there?
+            nnListing.put(child, dirStatus);
+          }
+        }
+      }
+      // Update the remaining count to include left mount points
+      if (nnListing.size() > 0) {
+        String lastListing = nnListing.lastKey();
+        for (int i = 0; i < children.size(); i++) {
+          if (children.get(i).compareTo(lastListing) > 0) {
+            remainingEntries += (children.size() - i);
+            break;
+          }
+        }
       }
     }
 
@@ -2108,6 +2132,36 @@ public class RouterClientProtocol implements ClientProtocol {
     }
   }
 
+  /**
+   * Check if we should add the mount point into the total listing.
+   * This should be done under either of the two cases:
+   * 1) current mount point is between startAfter and cutoff lastEntry.
+   * 2) there are no remaining entries from subclusters and this mount
+   *    point is bigger than all files from subclusters
+   * This is to make sure that the following batch of
+   * getListing call will use the correct startAfter, which is lastEntry from
+   * subcluster.
+   *
+   * @param mountPoint to be added mount point inside router
+   * @param lastEntry biggest listing from subcluster
+   * @param startAfter starting listing from client, used to define listing
+   *                   start boundary
+   * @param remainingEntries how many entries left from subcluster
+   * @return
+   */
+  private static boolean shouldAddMountPoint(
+      String mountPoint, String lastEntry, byte[] startAfter,
+      int remainingEntries) {
+    if (mountPoint.compareTo(DFSUtil.bytes2String(startAfter)) > 0 &&
+        mountPoint.compareTo(lastEntry) <= 0) {
+      return true;
+    }
+    if (remainingEntries == 0 && mountPoint.compareTo(lastEntry) >= 0) {
+      return true;
+    }
+    return false;
+  }
+
   /**
    * Checks if the path is a directory and is supposed to be present in all
    * subclusters.
diff --git a/hadoop-hdfs-project/hadoop-hdfs-rbf/src/test/java/org/apache/hadoop/hdfs/server/federation/MockResolver.java b/hadoop-hdfs-project/hadoop-hdfs-rbf/src/test/java/org/apache/hadoop/hdfs/server/federation/MockResolver.java
index 131dd74b8e0..39334250bc8 100644
--- a/hadoop-hdfs-project/hadoop-hdfs-rbf/src/test/java/org/apache/hadoop/hdfs/server/federation/MockResolver.java
+++ b/hadoop-hdfs-project/hadoop-hdfs-rbf/src/test/java/org/apache/hadoop/hdfs/server/federation/MockResolver.java
@@ -327,16 +327,30 @@ public class MockResolver
 
   @Override
   public List<String> getMountPoints(String path) throws IOException {
-    // Mounts only supported under root level
-    if (!path.equals("/")) {
-      return null;
-    }
     List<String> mounts = new ArrayList<>();
-    for (String mount : this.locations.keySet()) {
-      if (mount.length() > 1) {
-        // Remove leading slash, this is the behavior of the mount tree,
-        // return only names.
-        mounts.add(mount.replace("/", ""));
+    // for root path search, returning all downstream root level mapping
+    if (path.equals("/")) {
+      // Mounts only supported under root level
+      for (String mount : this.locations.keySet()) {
+        if (mount.length() > 1) {
+          // Remove leading slash, this is the behavior of the mount tree,
+          // return only names.
+          mounts.add(mount.replace("/", ""));
+        }
+      }
+    } else {
+      // a simplified version of MountTableResolver implementation
+      for (String key : this.locations.keySet()) {
+        if (key.startsWith(path)) {
+          String child = key.substring(path.length());
+          if (child.length() > 0) {
+            // only take children so remove parent path and /
+            mounts.add(key.substring(path.length()+1));
+          }
+        }
+      }
+      if (mounts.size() == 0) {
+        mounts = null;
       }
     }
     return mounts;
diff --git a/hadoop-hdfs-project/hadoop-hdfs-rbf/src/test/java/org/apache/hadoop/hdfs/server/federation/router/TestRouterRpc.java b/hadoop-hdfs-project/hadoop-hdfs-rbf/src/test/java/org/apache/hadoop/hdfs/server/federation/router/TestRouterRpc.java
index bfc712f057a..082094e5592 100644
--- a/hadoop-hdfs-project/hadoop-hdfs-rbf/src/test/java/org/apache/hadoop/hdfs/server/federation/router/TestRouterRpc.java
+++ b/hadoop-hdfs-project/hadoop-hdfs-rbf/src/test/java/org/apache/hadoop/hdfs/server/federation/router/TestRouterRpc.java
@@ -203,6 +203,9 @@ public class TestRouterRpc {
     cluster.addNamenodeOverrides(namenodeConf);
     cluster.setIndependentDNs();
 
+    Configuration conf = new Configuration();
+    conf.setInt(DFSConfigKeys.DFS_LIST_LIMIT, 5);
+    cluster.addNamenodeOverrides(conf);
     // Start NNs and DNs and wait until ready
     cluster.startCluster();
 
@@ -436,6 +439,62 @@ public class TestRouterRpc {
         new Object[] {badPath, HdfsFileStatus.EMPTY_NAME, false});
   }
 
+  @Test
+  public void testProxyListFilesLargeDir() throws IOException {
+    // Call listStatus against a dir with many files
+    // Create a parent point as well as a subfolder mount
+    // /parent
+    //    ns0 -> /parent
+    // /parent/file-7
+    //    ns0 -> /parent/file-7
+    // /parent/file-0
+    //    ns0 -> /parent/file-0
+    for (RouterContext rc : cluster.getRouters()) {
+      MockResolver resolver =
+          (MockResolver) rc.getRouter().getSubclusterResolver();
+      resolver.addLocation("/parent", ns, "/parent");
+      // file-0 is only in mount table
+      resolver.addLocation("/parent/file-0", ns, "/parent/file-0");
+      // file-7 is both in mount table and in file system
+      resolver.addLocation("/parent/file-7", ns, "/parent/file-7");
+    }
+
+    // Test the case when there is no subcluster path and only mount point
+    FileStatus[] result = routerFS.listStatus(new Path("/parent"));
+    assertEquals(2, result.length);
+    // this makes sure file[0-8] is added in order
+    assertEquals("file-0", result[0].getPath().getName());
+    assertEquals("file-7", result[1].getPath().getName());
+
+    // Create files and test full listing in order
+    NamenodeContext nn = cluster.getNamenode(ns, null);
+    FileSystem nnFileSystem = nn.getFileSystem();
+    for (int i = 1; i < 9; i++) {
+      createFile(nnFileSystem, "/parent/file-"+i, 32);
+    }
+
+    result = routerFS.listStatus(new Path("/parent"));
+    assertEquals(9, result.length);
+    // this makes sure file[0-8] is added in order
+    for (int i = 0; i < 9; i++) {
+      assertEquals("file-"+i, result[i].getPath().getName());
+    }
+
+    // Add file-9 and now this listing will be added from mount point
+    for (RouterContext rc : cluster.getRouters()) {
+      MockResolver resolver =
+          (MockResolver) rc.getRouter().getSubclusterResolver();
+      resolver.addLocation("/parent/file-9", ns, "/parent/file-9");
+    }
+    assertFalse(verifyFileExists(nnFileSystem, "/parent/file-9"));
+    result = routerFS.listStatus(new Path("/parent"));
+    // file-9 will be added by mount point
+    assertEquals(10, result.length);
+    for (int i = 0; i < 10; i++) {
+      assertEquals("file-"+i, result[i].getPath().getName());
+    }
+  }
+
   @Test
   public void testProxyListFilesWithConflict()
       throws IOException, InterruptedException {