HDFS-14353. Erasure Coding: metrics xmitsInProgress become to negative. Contributed by maobaolong.

Author: Inigo Goiri

hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/datanode/erasurecode/ErasureCodingWorker.java

@@ -170,4 +170,8 @@ public void shutDown() {
     stripedReconstructionPool.shutdown();
     stripedReadPool.shutdown();
   }
+
+  public float getXmitWeight() {
+    return xmitWeight;
+  }
 }

hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/datanode/erasurecode/StripedBlockReconstructor.java

@@ -67,7 +67,11 @@ public void run() {
       LOG.warn("Failed to reconstruct striped block: {}", getBlockGroup(), e);
       getDatanode().getMetrics().incrECFailedReconstructionTasks();
     } finally {
-      getDatanode().decrementXmitsInProgress(getXmits());
+      float xmitWeight = getErasureCodingWorker().getXmitWeight();
+      // if the xmits is smaller than 1, the xmitsSubmitted should be set to 1
+      // because if it set to zero, we cannot to measure the xmits submitted
+      int xmitsSubmitted = Math.max((int) (getXmits() * xmitWeight), 1);
+      getDatanode().decrementXmitsInProgress(xmitsSubmitted);
       final DataNodeMetrics metrics = getDatanode().getMetrics();
       metrics.incrECReconstructionTasks();
       metrics.incrECReconstructionBytesRead(getBytesRead());

hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/datanode/erasurecode/StripedReconstructor.java

@@ -275,4 +275,8 @@ Configuration getConf() {
   DataNode getDatanode() {
     return datanode;
   }
+
+  public ErasureCodingWorker getErasureCodingWorker() {
+    return erasureCodingWorker;
+  }
 }

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

@@ -514,6 +514,8 @@ private void testNNSendsErasureCodingTasks(int deadDN) throws Exception {
 
   @Test(timeout = 180000)
   public void testErasureCodingWorkerXmitsWeight() throws Exception {
+    testErasureCodingWorkerXmitsWeight(0.5f,
+        (int) (ecPolicy.getNumDataUnits() * 0.5f));
     testErasureCodingWorkerXmitsWeight(1f, ecPolicy.getNumDataUnits());
     testErasureCodingWorkerXmitsWeight(0f, 1);
     testErasureCodingWorkerXmitsWeight(10f, 10 * ecPolicy.getNumDataUnits());
@@ -567,6 +569,10 @@ public void stripedBlockReconstruction() throws IOException {
     } finally {
       barrier.await();
       DataNodeFaultInjector.set(oldInjector);
+      for (final DataNode curDn : cluster.getDataNodes()) {
+        GenericTestUtils.waitFor(() -> curDn.getXceiverCount() > 1, 10, 60000);
+        assertEquals(0, curDn.getXmitsInProgress());
+      }
     }
   }
 }