From 2da5aaab334d0d6a7dee244cac603aa35c9b0134 Mon Sep 17 00:00:00 2001
From: yliu <yliu@apache.org>
Date: Wed, 19 Aug 2015 21:55:43 +0800
Subject: [PATCH] HDFS-8908. TestAppendSnapshotTruncate may fail with
 IOException: Failed to replace a bad datanode. (Tsz Wo Nicholas Sze via yliu)

---
 hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt         |  3 +++
 .../hadoop/hdfs/TestAppendSnapshotTruncate.java     | 13 ++++++++-----
 2 files changed, 11 insertions(+), 5 deletions(-)

diff --git a/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt b/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt
index 038a6461457..f43f795599c 100644
--- a/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt
+++ b/hadoop-hdfs-project/hadoop-hdfs/CHANGES.txt
@@ -1174,6 +1174,9 @@ Release 2.8.0 - UNRELEASED
 
     HDFS-8891. HDFS concat should keep srcs order. (Yong Zhang via jing9)
 
+    HDFS-8908. TestAppendSnapshotTruncate may fail with IOException: Failed to
+    replace a bad datanode. (Tsz Wo Nicholas Sze via yliu)
+
 Release 2.7.2 - UNRELEASED
 
   INCOMPATIBLE CHANGES
diff --git a/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestAppendSnapshotTruncate.java b/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestAppendSnapshotTruncate.java
index d67ceb0f1f1..9a099870df3 100644
--- a/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestAppendSnapshotTruncate.java
+++ b/hadoop-hdfs-project/hadoop-hdfs/src/test/java/org/apache/hadoop/hdfs/TestAppendSnapshotTruncate.java
@@ -27,7 +27,6 @@ import java.io.RandomAccessFile;
 import java.util.Arrays;
 import java.util.HashMap;
 import java.util.Map;
-import java.util.Random;
 import java.util.concurrent.Callable;
 import java.util.concurrent.ThreadLocalRandom;
 import java.util.concurrent.atomic.AtomicBoolean;
@@ -42,6 +41,7 @@ import org.apache.hadoop.fs.FSDataOutputStream;
 import org.apache.hadoop.fs.FileStatus;
 import org.apache.hadoop.fs.FileUtil;
 import org.apache.hadoop.fs.Path;
+import org.apache.hadoop.hdfs.client.HdfsClientConfigKeys.BlockWrite.ReplaceDatanodeOnFailure;
 import org.apache.hadoop.hdfs.server.namenode.NameNode;
 import org.apache.hadoop.hdfs.server.namenode.TestFileTruncate;
 import org.apache.hadoop.test.GenericTestUtils;
@@ -64,10 +64,10 @@ public class TestAppendSnapshotTruncate {
   }
   private static final Log LOG = LogFactory.getLog(TestAppendSnapshotTruncate.class);
   private static final int BLOCK_SIZE = 1024;
-  private static final int DATANODE_NUM = 3;
+  private static final int DATANODE_NUM = 4;
   private static final short REPLICATION = 3;
-  private static final int FILE_WORKER_NUM = 3;
-  private static final long TEST_TIME_SECOND = 10;
+  private static final int FILE_WORKER_NUM = 10;
+  private static final long TEST_TIME_SECOND = 20;
   private static final long TEST_TIMEOUT_SECOND = TEST_TIME_SECOND + 60;
 
   static final int SHORT_HEARTBEAT = 1;
@@ -85,6 +85,7 @@ public class TestAppendSnapshotTruncate {
     conf.setInt(DFSConfigKeys.DFS_HEARTBEAT_INTERVAL_KEY, SHORT_HEARTBEAT);
     conf.setLong(
         DFSConfigKeys.DFS_NAMENODE_REPLICATION_PENDING_TIMEOUT_SEC_KEY, 1);
+    conf.setBoolean(ReplaceDatanodeOnFailure.BEST_EFFORT_KEY, true);
     cluster = new MiniDFSCluster.Builder(conf)
         .format(true)
         .numDataNodes(DATANODE_NUM)
@@ -476,7 +477,9 @@ public class TestAppendSnapshotTruncate {
     }
 
     void pause() {
-      Preconditions.checkState(state.compareAndSet(State.RUNNING, State.IDLE));
+      checkErrorState();
+      Preconditions.checkState(state.compareAndSet(State.RUNNING, State.IDLE),
+          "%s: state=%s != %s", name, state.get(), State.RUNNING);
     }
 
     void stop() throws InterruptedException {