Explorar o código

MAPREDUCE-4879. TeraOutputFormat may overwrite an existing output directory. (gera)

Gera Shegalov %!s(int64=10) %!d(string=hai) anos
pai
achega
25a0440238

+ 3 - 0
hadoop-mapreduce-project/CHANGES.txt

@@ -273,6 +273,9 @@ Release 2.7.0 - UNRELEASED
     MAPREDUCE-6160. Potential NullPointerException in MRClientProtocol
     interface implementation. (Rohith via jlowe)
 
+    MAPREDUCE-4879. TeraOutputFormat may overwrite an existing output
+    directory. (gera)
+
 Release 2.6.0 - 2014-11-18
 
   INCOMPATIBLE CHANGES

+ 0 - 4
hadoop-mapreduce-project/hadoop-mapreduce-examples/src/main/java/org/apache/hadoop/examples/terasort/TeraGen.java

@@ -289,10 +289,6 @@ public class TeraGen extends Configured implements Tool {
     }
     setNumberOfRows(job, parseHumanLong(args[0]));
     Path outputDir = new Path(args[1]);
-    if (outputDir.getFileSystem(getConf()).exists(outputDir)) {
-      throw new IOException("Output directory " + outputDir + 
-                            " already exists.");
-    }
     FileOutputFormat.setOutputPath(job, outputDir);
     job.setJobName("TeraGen");
     job.setJarByClass(TeraGen.class);

+ 26 - 1
hadoop-mapreduce-project/hadoop-mapreduce-examples/src/main/java/org/apache/hadoop/examples/terasort/TeraOutputFormat.java

@@ -20,10 +20,13 @@ package org.apache.hadoop.examples.terasort;
 
 import java.io.IOException;
 
+import org.apache.hadoop.conf.Configuration;
 import org.apache.hadoop.fs.FSDataOutputStream;
+import org.apache.hadoop.fs.FileStatus;
 import org.apache.hadoop.fs.FileSystem;
 import org.apache.hadoop.fs.Path;
 import org.apache.hadoop.io.Text;
+import org.apache.hadoop.mapred.FileAlreadyExistsException;
 import org.apache.hadoop.mapred.InvalidJobConfException;
 import org.apache.hadoop.mapreduce.JobContext;
 import org.apache.hadoop.mapreduce.OutputCommitter;
@@ -87,9 +90,31 @@ public class TeraOutputFormat extends FileOutputFormat<Text,Text> {
       throw new InvalidJobConfException("Output directory not set in JobConf.");
     }
 
+    final Configuration jobConf = job.getConfiguration();
+
     // get delegation token for outDir's file system
     TokenCache.obtainTokensForNamenodes(job.getCredentials(),
-        new Path[] { outDir }, job.getConfiguration());
+        new Path[] { outDir }, jobConf);
+
+    final FileSystem fs = outDir.getFileSystem(jobConf);
+
+    if (fs.exists(outDir)) {
+      // existing output dir is considered empty iff its only content is the
+      // partition file.
+      //
+      final FileStatus[] outDirKids = fs.listStatus(outDir);
+      boolean empty = false;
+      if (outDirKids != null && outDirKids.length == 1) {
+        final FileStatus st = outDirKids[0];
+        final String fname = st.getPath().getName();
+        empty =
+          !st.isDirectory() && TeraInputFormat.PARTITION_FILENAME.equals(fname);
+      }
+      if (TeraSort.getUseSimplePartitioner(job) || !empty) {
+        throw new FileAlreadyExistsException("Output directory " + outDir
+            + " already exists");
+      }
+    }
   }
 
   public RecordWriter<Text,Text> getRecordWriter(TaskAttemptContext job

+ 30 - 12
hadoop-mapreduce-project/hadoop-mapreduce-examples/src/test/java/org/apache/hadoop/examples/terasort/TestTeraSort.java

@@ -20,17 +20,19 @@ package org.apache.hadoop.examples.terasort;
 import java.io.File;
 import java.io.IOException;
 
+import org.apache.commons.logging.Log;
+import org.apache.commons.logging.LogFactory;
 import org.apache.hadoop.conf.Configuration;
 import org.apache.hadoop.fs.Path;
+import org.apache.hadoop.mapred.FileAlreadyExistsException;
 import org.apache.hadoop.mapred.HadoopTestCase;
 import org.apache.hadoop.util.ToolRunner;
-import org.junit.Ignore;
-@Ignore
 public class TestTeraSort extends HadoopTestCase {
+  private static Log LOG = LogFactory.getLog(TestTeraSort.class);
   
   public TestTeraSort()
       throws IOException {
-    super(CLUSTER_MR, DFS_FS, 1, 1);
+    super(LOCAL_MR, LOCAL_FS, 1, 1);
   }
 
   protected void tearDown() throws Exception {
@@ -45,42 +47,58 @@ public class TestTeraSort extends HadoopTestCase {
   private static final Path SORT_INPUT_PATH = new Path(TEST_DIR, "sortin");
   private static final Path SORT_OUTPUT_PATH = new Path(TEST_DIR, "sortout");
   private static final Path TERA_OUTPUT_PATH = new Path(TEST_DIR, "validate");
-  private static final String NUM_ROWS = "100"; 
+  private static final String NUM_ROWS = "100";
 
-  private void runTeraGen(Configuration conf, Path sortInput) 
+  private void runTeraGen(Configuration conf, Path sortInput)
       throws Exception {
     String[] genArgs = {NUM_ROWS, sortInput.toString()};
-    
+
     // Run TeraGen
     assertEquals(ToolRunner.run(conf, new TeraGen(), genArgs), 0);
   }
-  
+
   private void runTeraSort(Configuration conf,
       Path sortInput, Path sortOutput) throws Exception {
 
     // Setup command-line arguments to 'sort'
     String[] sortArgs = {sortInput.toString(), sortOutput.toString()};
-    
+
     // Run Sort
     assertEquals(ToolRunner.run(conf, new TeraSort(), sortArgs), 0);
   }
-  
-  private void runTeraValidator(Configuration job, 
-                                       Path sortOutput, Path valOutput) 
+
+  private void runTeraValidator(Configuration job,
+                                       Path sortOutput, Path valOutput)
   throws Exception {
     String[] svArgs = {sortOutput.toString(), valOutput.toString()};
 
     // Run Tera-Validator
     assertEquals(ToolRunner.run(job, new TeraValidate(), svArgs), 0);
   }
-  
+
   public void testTeraSort() throws Exception {
     // Run TeraGen to generate input for 'terasort'
     runTeraGen(createJobConf(), SORT_INPUT_PATH);
 
+    // Run teragen again to check for FAE
+    try {
+      runTeraGen(createJobConf(), SORT_INPUT_PATH);
+      fail("Teragen output overwritten!");
+    } catch (FileAlreadyExistsException fae) {
+      LOG.info("Expected exception: ", fae);
+    }
+
     // Run terasort
     runTeraSort(createJobConf(), SORT_INPUT_PATH, SORT_OUTPUT_PATH);
 
+    // Run terasort again to check for FAE
+    try {
+      runTeraSort(createJobConf(), SORT_INPUT_PATH, SORT_OUTPUT_PATH);
+      fail("Terasort output overwritten!");
+    } catch (FileAlreadyExistsException fae) {
+      LOG.info("Expected exception: ", fae);
+    }
+
     // Run tera-validator to check if sort worked correctly
     runTeraValidator(createJobConf(), SORT_OUTPUT_PATH,
       TERA_OUTPUT_PATH);