MAPREDUCE-5649. Reduce cannot use more than 2G memory for the final merge. Contributed by Gera Shegalov

(cherry picked from commit 7dc3c1203d)

(cherry picked from commit 87c2d915f1)

hadoop-mapreduce-project/CHANGES.txt

@@ -27,6 +27,9 @@ Release 2.7.2 - UNRELEASED
     MAPREDUCE-6474. ShuffleHandler can possibly exhaust nodemanager file
     descriptors (Kuhu Shukla via jlowe)
 
+    MAPREDUCE-5649. Reduce cannot use more than 2G memory for the final merge
+    (Gera Shegalov via jlowe)
+
 Release 2.7.1 - 2015-07-06 
 
   INCOMPATIBLE CHANGES

hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core/src/main/java/org/apache/hadoop/mapreduce/task/reduce/MergeManagerImpl.java

@@ -94,7 +94,9 @@ public class MergeManagerImpl<K, V> implements MergeManager<K, V> {
   Set<CompressAwarePath> onDiskMapOutputs = new TreeSet<CompressAwarePath>();
   private final OnDiskMerger onDiskMerger;
 
-  private final long memoryLimit;
+  @VisibleForTesting
+  final long memoryLimit;
+
   private long usedMemory;
   private long commitMemory;
   private final long maxSingleShuffleLimit;
@@ -167,10 +169,9 @@ public class MergeManagerImpl<K, V> implements MergeManager<K, V> {
     }
 
     // Allow unit tests to fix Runtime memory
-    this.memoryLimit = 
-      (long)(jobConf.getLong(MRJobConfig.REDUCE_MEMORY_TOTAL_BYTES,
-          Math.min(Runtime.getRuntime().maxMemory(), Integer.MAX_VALUE))
-        * maxInMemCopyUse);
+    this.memoryLimit = (long)(jobConf.getLong(
+        MRJobConfig.REDUCE_MEMORY_TOTAL_BYTES,
+        Runtime.getRuntime().maxMemory()) * maxInMemCopyUse);
 
     this.ioSortFactor = jobConf.getInt(MRJobConfig.IO_SORT_FACTOR, 100);
 
@@ -667,6 +668,18 @@ public class MergeManagerImpl<K, V> implements MergeManager<K, V> {
     }
   }
 
+  @VisibleForTesting
+  final long getMaxInMemReduceLimit() {
+    final float maxRedPer =
+        jobConf.getFloat(MRJobConfig.REDUCE_INPUT_BUFFER_PERCENT, 0f);
+    if (maxRedPer > 1.0 || maxRedPer < 0.0) {
+      throw new RuntimeException(maxRedPer + ": "
+          + MRJobConfig.REDUCE_INPUT_BUFFER_PERCENT
+          + " must be a float between 0 and 1.0");
+    }
+    return (long)(memoryLimit * maxRedPer);
+  }
+
   private RawKeyValueIterator finalMerge(JobConf job, FileSystem fs,
                                        List<InMemoryMapOutput<K,V>> inMemoryMapOutputs,
                                        List<CompressAwarePath> onDiskMapOutputs
@@ -674,17 +687,7 @@ public class MergeManagerImpl<K, V> implements MergeManager<K, V> {
     LOG.info("finalMerge called with " +
         inMemoryMapOutputs.size() + " in-memory map-outputs and " +
         onDiskMapOutputs.size() + " on-disk map-outputs");
-    
-    final float maxRedPer =
-      job.getFloat(MRJobConfig.REDUCE_INPUT_BUFFER_PERCENT, 0f);
-    if (maxRedPer > 1.0 || maxRedPer < 0.0) {
-      throw new IOException(MRJobConfig.REDUCE_INPUT_BUFFER_PERCENT +
-                            maxRedPer);
-    }
-    int maxInMemReduce = (int)Math.min(
-        Runtime.getRuntime().maxMemory() * maxRedPer, Integer.MAX_VALUE);
-    
-
+    final long maxInMemReduce = getMaxInMemReduceLimit();
     // merge config params
     Class<K> keyClass = (Class<K>)job.getMapOutputKeyClass();
     Class<V> valueClass = (Class<V>)job.getMapOutputValueClass();

hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core/src/test/java/org/apache/hadoop/mapreduce/task/reduce/TestMergeManager.java

@@ -260,4 +260,33 @@ public class TestMergeManager {
     }
 
   }
+
+  @Test
+  public void testLargeMemoryLimits() throws Exception {
+    final JobConf conf = new JobConf();
+    // Xmx in production
+    conf.setLong(MRJobConfig.REDUCE_MEMORY_TOTAL_BYTES,
+        8L * 1024 * 1024 * 1024);
+
+    // M1 = Xmx fraction for map outputs
+    conf.setFloat(MRJobConfig.SHUFFLE_INPUT_BUFFER_PERCENT, 1.0f);
+
+    // M2 = max M1 fraction for a single maple output
+    conf.setFloat(MRJobConfig.SHUFFLE_MEMORY_LIMIT_PERCENT, 0.95f);
+
+    // M3 = M1 fraction at which in memory merge is triggered
+    conf.setFloat(MRJobConfig.SHUFFLE_MERGE_PERCENT, 1.0f);
+
+    // M4 = M1 fraction of map outputs remaining in memory for a reduce
+    conf.setFloat(MRJobConfig.REDUCE_INPUT_BUFFER_PERCENT, 1.0f);
+
+    final MergeManagerImpl<Text, Text> mgr = new MergeManagerImpl<Text, Text>(
+        null, conf, mock(LocalFileSystem.class), null, null, null, null, null,
+        null, null, null, null, null, new MROutputFiles());
+    assertTrue("Large shuffle area unusable: " + mgr.memoryLimit,
+        mgr.memoryLimit > Integer.MAX_VALUE);
+    final long maxInMemReduce = mgr.getMaxInMemReduceLimit();
+    assertTrue("Large in-memory reduce area unusable: " + maxInMemReduce,
+        maxInMemReduce > Integer.MAX_VALUE);
+  }
 }