From 0e6604aab30671618355982aa86c0a7661132a9c Mon Sep 17 00:00:00 2001
From: Sanjay Radia <sradia@apache.org>
Date: Wed, 27 Mar 2013 20:49:48 +0000
Subject: [PATCH] HDFS-4633 TestDFSClientExcludedNodes fails sporadically if
 excluded nodes cache expires too quickly  (Chris Nauroth via Sanjay)

git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/trunk@1461846 13f79535-47bb-0310-9956-ffa450edef68
---
 hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt   |  3 ++
 .../hdfs/TestDFSClientExcludedNodes.java      | 38 +++++++++++++------
 2 files changed, 30 insertions(+), 11 deletions(-)

diff --git a/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt b/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt
index 6b2b2aed79..7a9e504c43 100644
--- a/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt
+++ b/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt
@@ -181,6 +181,9 @@ Trunk (Unreleased)
     HDFS-4346. Add SequentialNumber as a base class for INodeId and
     GenerationStamp.  (szetszwo)
 
+    HDFS-4633 TestDFSClientExcludedNodes fails sporadically if excluded nodes
+    cache expires too quickly  (Chris Nauroth via Sanjay)
+
   OPTIMIZATIONS
 
   BUG FIXES
diff --git a/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestDFSClientExcludedNodes.java b/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestDFSClientExcludedNodes.java
index fa88e39457..3574323d54 100644
--- a/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestDFSClientExcludedNodes.java
+++ b/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestDFSClientExcludedNodes.java
@@ -32,6 +32,8 @@
 import org.apache.hadoop.hdfs.MiniDFSCluster.DataNodeProperties;
 import org.apache.hadoop.util.ThreadUtil;
 
+import org.junit.After;
+import org.junit.Before;
 import org.junit.Test;
 
 
@@ -41,10 +43,25 @@
  */
 public class TestDFSClientExcludedNodes {
 
-  @Test(timeout=10000)
+  private MiniDFSCluster cluster;
+  private Configuration conf;
+
+  @Before
+  public void setUp() {
+    cluster = null;
+    conf = new HdfsConfiguration();
+  }
+
+  @After
+  public void tearDown() {
+    if (cluster != null) {
+      cluster.shutdown();
+    }
+  }
+
+  @Test(timeout=60000)
   public void testExcludedNodes() throws IOException {
-    Configuration conf = new HdfsConfiguration();
-    MiniDFSCluster cluster = new MiniDFSCluster.Builder(conf).numDataNodes(3).build();
+    cluster = new MiniDFSCluster.Builder(conf).numDataNodes(3).build();
     FileSystem fs = cluster.getFileSystem();
     Path filePath = new Path("/testExcludedNodes");
 
@@ -67,17 +84,16 @@ public void testExcludedNodes() throws IOException {
     }
   }
 
-  @Test(timeout=10000)
+  @Test(timeout=60000)
   public void testExcludedNodesForgiveness() throws IOException {
-    Configuration conf = new HdfsConfiguration();
-    // Forgive nodes in under 1s for this test case.
+    // Forgive nodes in under 2.5s for this test case.
     conf.setLong(
         DFSConfigKeys.DFS_CLIENT_WRITE_EXCLUDE_NODES_CACHE_EXPIRY_INTERVAL,
-        1000);
+        2500);
     // We'll be using a 512 bytes block size just for tests
     // so making sure the checksum bytes too match it.
     conf.setInt("io.bytes.per.checksum", 512);
-    MiniDFSCluster cluster = new MiniDFSCluster.Builder(conf).numDataNodes(3).build();
+    cluster = new MiniDFSCluster.Builder(conf).numDataNodes(3).build();
     List<DataNodeProperties> props = cluster.dataNodes;
     FileSystem fs = cluster.getFileSystem();
     Path filePath = new Path("/testForgivingExcludedNodes");
@@ -112,11 +128,11 @@ public void testExcludedNodesForgiveness() throws IOException {
     Assert.assertEquals(true, cluster.restartDataNode(two, true));
     cluster.waitActive();
 
-    // Sleep for 2s, to let the excluded nodes be expired
+    // Sleep for 5s, to let the excluded nodes be expired
     // from the excludes list (i.e. forgiven after the configured wait period).
-    // [Sleeping just in case the restart of the DNs completed < 2s cause
+    // [Sleeping just in case the restart of the DNs completed < 5s cause
     // otherwise, we'll end up quickly excluding those again.]
-    ThreadUtil.sleepAtLeastIgnoreInterrupts(2000);
+    ThreadUtil.sleepAtLeastIgnoreInterrupts(5000);
 
     // Terminate the last good DN, to assert that there's no
     // single-DN-available scenario, caused by not forgiving the other