HADOOP-5620. Add an option to DistCp for preserving modification and access times. Contributed by Rodrigo Schmidt

git-svn-id: https://svn.apache.org/repos/asf/hadoop/core/trunk@779796 13f79535-47bb-0310-9956-ffa450edef68

CHANGES.txt

@@ -390,7 +390,7 @@ Trunk (unreleased changes)
     TestNameEditsConfig, TestStartup and TestStorageRestore.
     (Jakob Homan via shv)
 
-    HADOOP-5438. Provide a single FileSystem method to create or open-for-append 
+    HADOOP-5438. Provide a single FileSystem method to create or open-for-append
     to a file.  (He Yongqiang via dhruba)
 
     HADOOP-5472. Change DistCp to support globbing of input paths.  (Dhruba
@@ -398,6 +398,9 @@ Trunk (unreleased changes)
 
     HADOOP-5175. Don't unpack libjars on classpath. (Todd Lipcon via tomwhite)
 
+    HADOOP-5620. Add an option to DistCp for preserving modification and access
+    times.  (Rodrigo Schmidt via szetszwo)
+
   OPTIMIZATIONS
 
     HADOOP-5595. NameNode does not need to run a replicator to choose a

src/test/hdfs-with-mr/org/apache/hadoop/fs/TestCopyFiles.java

@@ -380,7 +380,7 @@ public class TestCopyFiles extends TestCase {
         deldir(hdfs, "/logs");
 
         ToolRunner.run(new DistCp(conf), new String[] {
-                                         "-p",
+                                         "-prbugp", // no t to avoid preserving mod. times
                                          "-update",
                                          "-log",
                                          namenode+"/logs",
@@ -393,7 +393,7 @@ public class TestCopyFiles extends TestCase {
 
         deldir(hdfs, "/logs");
         ToolRunner.run(new DistCp(conf), new String[] {
-                                         "-p",
+                                         "-prbugp", // no t to avoid preserving mod. times
                                          "-overwrite",
                                          "-log",
                                          namenode+"/logs",
@@ -551,6 +551,32 @@ public class TestCopyFiles extends TestCase {
         deldir(fs, "/destdat");
         deldir(fs, "/srcdat");
       }
+
+      {//test preserving times
+        MyFile[] files = createFiles(URI.create(nnUri), "/srcdat");
+        fs.mkdirs(new Path("/srcdat/tmpf1"));
+        fs.mkdirs(new Path("/srcdat/tmpf2"));
+        FileStatus[] srcstat = getFileStatus(fs, "/srcdat", files);
+        FsPermission[] permissions = new FsPermission[srcstat.length];
+        for(int i = 0; i < srcstat.length; i++) {
+          fs.setTimes(srcstat[i].getPath(), 40, 50);
+        }
+
+        ToolRunner.run(new DistCp(conf),
+            new String[]{"-pt", nnUri+"/srcdat", nnUri+"/destdat"});
+
+        FileStatus[] dststat = getFileStatus(fs, "/destdat", files);
+        for(int i = 0; i < dststat.length; i++) {
+          assertEquals("Modif. Time i=" + i, 40, dststat[i].getModificationTime());
+          assertEquals("Access Time i=" + i+ srcstat[i].getPath() + "-" + dststat[i].getPath(), 50, dststat[i].getAccessTime());
+        }
+        
+        assertTrue("Source and destination directories do not match.",
+                   checkFiles(fs, "/destdat", files));
+  
+        deldir(fs, "/destdat");
+        deldir(fs, "/srcdat");
+      }
     } finally {
       if (cluster != null) { cluster.shutdown(); }
     }

src/tools/org/apache/hadoop/tools/DistCp.java

@@ -86,7 +86,8 @@ public class DistCp implements Tool {
     "\n                       u: user" + 
     "\n                       g: group" +
     "\n                       p: permission" +
-    "\n                       -p alone is equivalent to -prbugp" +
+    "\n                       t: modification and access times" +
+    "\n                       -p alone is equivalent to -prbugpt" +
     "\n-i                     Ignore failures" +
     "\n-log <logdir>          Write logs to <logdir>" +
     "\n-m <num_maps>          Maximum number of simultaneous copies" +
@@ -146,7 +147,7 @@ public class DistCp implements Tool {
     }
   }
   static enum FileAttribute {
-    BLOCK_SIZE, REPLICATION, USER, GROUP, PERMISSION;
+    BLOCK_SIZE, REPLICATION, USER, GROUP, PERMISSION, TIMES;
 
     final char symbol;
 
@@ -460,7 +461,7 @@ public class DistCp implements Tool {
               + ") but expected " + bytesString(srcstat.getLen()) 
               + " from " + srcstat.getPath());        
         } 
-        updatePermissions(srcstat, dststat);
+        updateDestStatus(srcstat, dststat);
       }
 
       // report at least once for each file
@@ -486,10 +487,10 @@ public class DistCp implements Tool {
       }
     }
 
-    private void updatePermissions(FileStatus src, FileStatus dst
+    private void updateDestStatus(FileStatus src, FileStatus dst
         ) throws IOException {
       if (preserve_status) {
-        DistCp.updatePermissions(src, dst, preseved, destFileSys);
+        DistCp.updateDestStatus(src, dst, preseved, destFileSys);
       }
     }
 
@@ -669,7 +670,7 @@ public class DistCp implements Tool {
     }
   }
 
-  private static void updatePermissions(FileStatus src, FileStatus dst,
+  private static void updateDestStatus(FileStatus src, FileStatus dst,
       EnumSet<FileAttribute> preseved, FileSystem destFileSys
       ) throws IOException {
     String owner = null;
@@ -689,6 +690,9 @@ public class DistCp implements Tool {
         && !src.getPermission().equals(dst.getPermission())) {
       destFileSys.setPermission(dst.getPath(), src.getPermission());
     }
+    if (preseved.contains(FileAttribute.TIMES)) {
+      destFileSys.setTimes(dst.getPath(), src.getModificationTime(), src.getAccessTime());
+    }
   }
 
   static private void finalize(Configuration conf, JobConf jobconf,
@@ -713,7 +717,7 @@ public class DistCp implements Tool {
       FilePair pair = new FilePair(); 
       for(; in.next(dsttext, pair); ) {
         Path absdst = new Path(destPath, pair.output);
-        updatePermissions(pair.input, dstfs.getFileStatus(absdst),
+        updateDestStatus(pair.input, dstfs.getFileStatus(absdst),
             preseved, dstfs);
       }
     } finally {