HBASE-5747 Forward port "hbase-5708 [89-fb] Make MiniMapRedCluster directory a subdirectory of target/test"

git-svn-id: https://svn.apache.org/repos/asf/hbase/trunk@1326000 13f79535-47bb-0310-9956-ffa450edef68

src/main/java/org/apache/hadoop/hbase/HConstants.java

@@ -103,6 +103,10 @@ public final class HConstants {
   /** by default every master is a possible primary master unless the conf explicitly overrides it */
   public static final boolean DEFAULT_MASTER_TYPE_BACKUP = false;
 
+  /** Parameter name for ZooKeeper session time out.*/
+  public static final String ZOOKEEPER_SESSION_TIMEOUT =
+    "zookeeper.session.timeout";
+
   /** Name of ZooKeeper quorum configuration parameter. */
   public static final String ZOOKEEPER_QUORUM = "hbase.zookeeper.quorum";
 
@@ -655,6 +659,8 @@ public final class HConstants {
 /** Region in Transition metrics threshold time */
   public static final String METRICS_RIT_STUCK_WARNING_THRESHOLD="hbase.metrics.rit.stuck.warning.threshold";
 
+  public static final String LOAD_BALANCER_SLOP_KEY = "hbase.regions.slop";
+
   private HConstants() {
     // Can't be instantiated with this ctor.
   }

src/main/java/org/apache/hadoop/hbase/master/SplitLogManager.java

@@ -1004,6 +1004,10 @@ public class SplitLogManager extends ZooKeeperListener {
         Stat stat) {
       tot_mgr_get_data_result.incrementAndGet();
       if (rc != 0) {
+        if (rc == KeeperException.Code.SESSIONEXPIRED.intValue()) {
+          LOG.error("ZK session expired. Master is expected to shut down. Abandoning retries.");
+          return;
+        }
         if (rc == KeeperException.Code.NONODE.intValue()) {
           tot_mgr_get_data_nonode.incrementAndGet();
           // The task znode has been deleted. Must be some pending delete
@@ -1085,6 +1089,10 @@ public class SplitLogManager extends ZooKeeperListener {
     @Override
     public void processResult(int rc, String path, Object ctx, String name) {
       if (rc != 0) {
+        if (rc == KeeperException.Code.SESSIONEXPIRED.intValue()) {
+          LOG.error("ZK session expired. Master is expected to shut down. Abandoning retries.");
+          return;
+        }
         Long retry_count = (Long)ctx;
         LOG.warn("rc=" + KeeperException.Code.get(rc) + " for "+ path +
             " remaining retries=" + retry_count);

src/test/java/org/apache/hadoop/hbase/HBaseTestCase.java

@@ -19,7 +19,6 @@
  */
 package org.apache.hadoop.hbase;
 
-import java.io.File;
 import java.io.IOException;
 import java.io.UnsupportedEncodingException;
 import java.util.Iterator;
@@ -54,16 +53,6 @@ import org.apache.hadoop.hdfs.MiniDFSCluster;
 public abstract class HBaseTestCase extends TestCase {
   private static final Log LOG = LogFactory.getLog(HBaseTestCase.class);
 
-  /** configuration parameter name for test directory
-   * @deprecated see HBaseTestingUtility#TEST_DIRECTORY_KEY
-   **/
-  private static final String TEST_DIRECTORY_KEY = "test.build.data";
-
-/*
-  protected final static byte [] fam1 = Bytes.toBytes("colfamily1");
-  protected final static byte [] fam2 = Bytes.toBytes("colfamily2");
-  protected final static byte [] fam3 = Bytes.toBytes("colfamily3");
-*/
   protected final static byte [] fam1 = Bytes.toBytes("colfamily11");
   protected final static byte [] fam2 = Bytes.toBytes("colfamily21");
   protected final static byte [] fam3 = Bytes.toBytes("colfamily31");
@@ -82,9 +71,7 @@ public abstract class HBaseTestCase extends TestCase {
   protected String START_KEY;
   protected static final int MAXVERSIONS = 3;
 
-  static {
-    initialize();
-  }
+  protected final HBaseTestingUtility testUtil = new HBaseTestingUtility();
 
   public volatile Configuration conf;
 
@@ -161,19 +148,12 @@ public abstract class HBaseTestCase extends TestCase {
    * @return directory to use for this test
    */
     protected Path getUnitTestdir(String testName) {
-      return new Path(
-          System.getProperty(
-            HBaseTestingUtility.BASE_TEST_DIRECTORY_KEY,
-            HBaseTestingUtility.DEFAULT_BASE_TEST_DIRECTORY
-            ),
-        testName
-      );
+      return testUtil.getDataTestDir(testName);
     }
 
   protected HRegion createNewHRegion(HTableDescriptor desc, byte [] startKey,
       byte [] endKey)
   throws IOException {
-    FileSystem filesystem = FileSystem.get(conf);
     HRegionInfo hri = new HRegionInfo(desc.getName(), startKey, endKey);
     return HRegion.createHRegion(hri, testDir, conf, desc);
   }
@@ -628,21 +608,6 @@ public abstract class HBaseTestCase extends TestCase {
       }
     }
 
-  /**
-   * Initializes parameters used in the test environment:
-   *
-   * Sets the configuration parameter TEST_DIRECTORY_KEY if not already set.
-   * Sets the boolean debugging if "DEBUGGING" is set in the environment.
-   * If debugging is enabled, reconfigures logging so that the root log level is
-   * set to WARN and the logging level for the package is set to DEBUG.
-   */
-  public static void initialize() {
-    if (System.getProperty(TEST_DIRECTORY_KEY) == null) {
-      System.setProperty(TEST_DIRECTORY_KEY, new File(
-          "build/hbase/test").getAbsolutePath());
-    }
-  }
-
   /**
    * Common method to close down a MiniDFSCluster and the associated file system
    *

src/test/java/org/apache/hadoop/hbase/HBaseTestingUtility.java

@@ -25,6 +25,7 @@ import java.io.File;
 import java.io.IOException;
 import java.io.OutputStream;
 import java.lang.reflect.Field;
+import java.lang.reflect.Modifier;
 import java.net.InetAddress;
 import java.net.ServerSocket;
 import java.net.Socket;
@@ -84,8 +85,8 @@ import org.apache.hadoop.hbase.zookeeper.ZooKeeperWatcher;
 import org.apache.hadoop.hdfs.DFSClient;
 import org.apache.hadoop.hdfs.DistributedFileSystem;
 import org.apache.hadoop.hdfs.MiniDFSCluster;
-import org.apache.hadoop.mapred.JobConf;
 import org.apache.hadoop.mapred.MiniMRCluster;
+import org.apache.hadoop.mapred.TaskLog;
 import org.apache.zookeeper.KeeperException;
 import org.apache.zookeeper.KeeperException.NodeExistsException;
 import org.apache.zookeeper.WatchedEvent;
@@ -124,11 +125,15 @@ public class HBaseTestingUtility {
   private MiniHBaseCluster hbaseCluster = null;
   private MiniMRCluster mrCluster = null;
 
-  // Directory where we put the data for this instance of HBaseTestingUtility
+  /** If there is a mini cluster running for this testing utility instance. */
+  private boolean miniClusterRunning;
+
+  private String hadoopLogDir;
+
+  // Directory where we put the data for this instance of HBaseTestingUtility.
   private File dataTestDir = null;
 
-  // Directory (usually a subdirectory of dataTestDir) used by the dfs cluster
-  //  if any
+  // Directory (a subdirectory of dataTestDir) used by the dfs cluster if any
   private File clusterTestDir = null;
 
   /**
@@ -151,6 +156,9 @@ public class HBaseTestingUtility {
    */
   public static final String DEFAULT_BASE_TEST_DIRECTORY = "target/test-data";
 
+  /** Filesystem URI used for map-reduce mini-cluster setup */
+  private static String FS_URI;
+
   /** Compression algorithms to use in parameterized JUnit 4 tests */
   public static final List<Object[]> COMPRESSION_ALGORITHMS_PARAMETERIZED =
     Arrays.asList(new Object[][] {
@@ -236,24 +244,42 @@ public class HBaseTestingUtility {
    * @see #getTestFileSystem()
    */
   public Path getDataTestDir() {
-    if (dataTestDir == null){
+    if (this.dataTestDir == null){
       setupDataTestDir();
     }
-    return new Path(dataTestDir.getAbsolutePath());
+    return new Path(this.dataTestDir.getAbsolutePath());
   }
 
   /**
    * @return Where the DFS cluster will write data on the local subsystem.
-   * Creates it if it does not exist already.
+   * Creates it if it does not exist already.  A subdir of {@link #getBaseTestDir()}
    * @see #getTestFileSystem()
    */
-  public Path getClusterTestDir() {
+  Path getClusterTestDir() {
     if (clusterTestDir == null){
       setupClusterTestDir();
     }
     return new Path(clusterTestDir.getAbsolutePath());
   }
 
+  /**
+   * Creates a directory for the DFS cluster, under the test data
+   */
+  private void setupClusterTestDir() {
+    if (clusterTestDir != null) {
+      return;
+    }
+
+    // Using randomUUID ensures that multiple clusters can be launched by
+    //  a same test, if it stops & starts them
+    Path testDir = getDataTestDir("dfscluster_" + UUID.randomUUID().toString());
+    clusterTestDir = new File(testDir.toString()).getAbsoluteFile();
+    // Have it cleaned up on exit
+    clusterTestDir.deleteOnExit();
+    conf.set(TEST_DIRECTORY_KEY, clusterTestDir.getPath());
+    LOG.info("Created new mini-cluster data directory: " + clusterTestDir);
+  }
+
   /**
    * @param subdirName
    * @return Path to a subdirectory named <code>subdirName</code> under
@@ -281,7 +307,7 @@ public class HBaseTestingUtility {
    * @return The calculated data test build directory.
    */
   private void setupDataTestDir() {
-    if (dataTestDir != null) {
+    if (this.dataTestDir != null) {
       LOG.warn("Data test dir already setup in " +
         dataTestDir.getAbsolutePath());
       return;
@@ -290,8 +316,8 @@ public class HBaseTestingUtility {
     String randomStr = UUID.randomUUID().toString();
     Path testPath= new Path(getBaseTestDir(), randomStr);
 
-    dataTestDir = new File(testPath.toString()).getAbsoluteFile();
-    dataTestDir.deleteOnExit();
+    this.dataTestDir = new File(testPath.toString()).getAbsoluteFile();
+    this.dataTestDir.deleteOnExit();
 
     createSubDirAndSystemProperty(
       "hadoop.log.dir",
@@ -329,7 +355,7 @@ public class HBaseTestingUtility {
       // There is already a value set. So we do nothing but hope
       //  that there will be no conflicts
       LOG.info("System.getProperty(\""+propertyName+"\") already set to: "+
-        sysValue + " so I do NOT create it in "+dataTestDir.getAbsolutePath());
+        sysValue + " so I do NOT create it in " + this.dataTestDir.getAbsolutePath());
       String confValue = conf.get(propertyName);
       if (confValue != null && !confValue.endsWith(sysValue)){
        LOG.warn(
@@ -346,33 +372,6 @@ public class HBaseTestingUtility {
     }
   }
 
-  /**
-   * Creates a directory for the DFS cluster, under the test data
-   */
-  private void setupClusterTestDir() {
-    if (clusterTestDir != null) {
-      LOG.warn("Cluster test dir already setup in " +
-        clusterTestDir.getAbsolutePath());
-      return;
-    }
-
-    // Using randomUUID ensures that multiple clusters can be launched by
-    //  a same test, if it stops & starts them
-    Path testDir = getDataTestDir("dfscluster_" + UUID.randomUUID().toString());
-    clusterTestDir = new File(testDir.toString()).getAbsoluteFile();
-    // Have it cleaned up on exit
-    clusterTestDir.deleteOnExit();
-  }
-
-  /**
-   * @throws IOException If a cluster -- zk, dfs, or hbase -- already running.
-   */
-  public void isRunningCluster() throws IOException {
-    if (dfsCluster == null) return;
-    throw new IOException("Cluster already running at " +
-      this.clusterTestDir);
-  }
-
   /**
    * Start a minidfscluster.
    * @param servers How many DNs to start.
@@ -396,7 +395,7 @@ public class HBaseTestingUtility {
    * @return The mini dfs cluster created.
    */
   public MiniDFSCluster startMiniDFSCluster(final String hosts[])
-    throws Exception {
+  throws Exception {
     if ( hosts != null && hosts.length != 0) {
       return startMiniDFSCluster(hosts.length, hosts);
     } else {
@@ -415,24 +414,7 @@ public class HBaseTestingUtility {
    */
   public MiniDFSCluster startMiniDFSCluster(int servers, final String hosts[])
   throws Exception {
-
-    // Check that there is not already a cluster running
-    isRunningCluster();
-
-    // Initialize the local directory used by the MiniDFS
-    if (clusterTestDir == null) {
-      setupClusterTestDir();
-    }
-
-    // We have to set this property as it is used by MiniCluster
-    System.setProperty(TEST_DIRECTORY_KEY, this.clusterTestDir.toString());
-
-    // Some tests also do this:
-    //  System.getProperty("test.cache.data", "build/test/cache");
-    // It's also deprecated
-    System.setProperty("test.cache.data", this.clusterTestDir.toString());
-
-    // Ok, now we can start
+    createDirsAndSetProperties();
     this.dfsCluster = new MiniDFSCluster(0, this.conf, servers, true, true,
       true, null, null, hosts, null);
 
@@ -448,18 +430,46 @@ public class HBaseTestingUtility {
     return this.dfsCluster;
   }
 
+  public MiniDFSCluster startMiniDFSClusterForTestHLog(int namenodePort) throws IOException {
+    createDirsAndSetProperties();
+    dfsCluster = new MiniDFSCluster(namenodePort, conf, 5, false, true, true, null,
+        null, null, null);
+    return dfsCluster;
+  }
+
+  /** This is used before starting HDFS and map-reduce mini-clusters */
+  private void createDirsAndSetProperties() {
+    setupClusterTestDir();
+    System.setProperty(TEST_DIRECTORY_KEY, clusterTestDir.getPath());
+    createDirAndSetProperty("cache_data", "test.cache.data");
+    createDirAndSetProperty("hadoop_tmp", "hadoop.tmp.dir");
+    hadoopLogDir = createDirAndSetProperty("hadoop_logs", "hadoop.log.dir");
+    createDirAndSetProperty("mapred_output", "mapred.output.dir");
+    createDirAndSetProperty("mapred_local", "mapred.local.dir");
+    createDirAndSetProperty("mapred_system", "mapred.system.dir");
+    createDirAndSetProperty("mapred_temp", "mapred.temp.dir");
+  }
+
+  private String createDirAndSetProperty(final String relPath, String property) {
+    String path = clusterTestDir.getPath() + "/" + relPath;
+    System.setProperty(property, path);
+    conf.set(property, path);
+    new File(path).mkdirs();
+    LOG.info("Setting " + property + " to " + path + " in system properties and HBase conf");
+    return path;
+  }
+
   /**
    * Shuts down instance created by call to {@link #startMiniDFSCluster(int)}
    * or does nothing.
-   * @throws Exception
+   * @throws IOException
    */
-  public void shutdownMiniDFSCluster() throws Exception {
+  public void shutdownMiniDFSCluster() throws IOException {
     if (this.dfsCluster != null) {
       // The below throws an exception per dn, AsynchronousCloseException.
       this.dfsCluster.shutdown();
       dfsCluster = null;
     }
-
   }
 
   /**
@@ -483,8 +493,8 @@ public class HBaseTestingUtility {
    */
   public MiniZooKeeperCluster startMiniZKCluster(int zooKeeperServerNum)
       throws Exception {
-    File zkClusterFile = new File(getClusterTestDir().toString());
-    return startMiniZKCluster(zkClusterFile, zooKeeperServerNum);
+    setupClusterTestDir();
+    return startMiniZKCluster(clusterTestDir, zooKeeperServerNum);
   }
 
   private MiniZooKeeperCluster startMiniZKCluster(final File dir)
@@ -597,7 +607,13 @@ public class HBaseTestingUtility {
         numSlaves + " regionserver(s) and " + numDataNodes + " datanode(s)");
 
     // If we already put up a cluster, fail.
-    isRunningCluster();
+    if (miniClusterRunning) {
+      throw new IllegalStateException("A mini-cluster is already running");
+    }
+    miniClusterRunning = true;
+
+    setupClusterTestDir();
+    System.setProperty(TEST_DIRECTORY_KEY, this.clusterTestDir.getPath());
 
     // Bring up mini dfs cluster. This spews a bunch of warnings about missing
     // scheme. Complaints are 'Scheme is undefined for build/test/data/dfs/name1'.
@@ -691,15 +707,8 @@ public class HBaseTestingUtility {
     }
     shutdownMiniDFSCluster();
 
-    // Clean up our directory.
-    if (this.clusterTestDir != null && this.clusterTestDir.exists()) {
-      // Need to use deleteDirectory because File.delete required dir is empty.
-      if (!FSUtils.deleteDirectory(FileSystem.getLocal(this.conf),
-          new Path(this.clusterTestDir.toString()))) {
-        LOG.warn("Failed delete of " + this.clusterTestDir.toString());
-      }
-      this.clusterTestDir = null;
-    }
+    cleanupTestDir();
+    miniClusterRunning = false;
     LOG.info("Minicluster is down");
   }
 
@@ -1265,28 +1274,59 @@ public class HBaseTestingUtility {
    *
    * @throws IOException When starting the cluster fails.
    */
-  public void startMiniMapReduceCluster() throws IOException {
+  public MiniMRCluster startMiniMapReduceCluster() throws IOException {
     startMiniMapReduceCluster(2);
+    return mrCluster;
   }
 
   /**
-   * Starts a <code>MiniMRCluster</code>.
-   *
+   * Tasktracker has a bug where changing the hadoop.log.dir system property
+   * will not change its internal static LOG_DIR variable.
+   */
+  private void forceChangeTaskLogDir() {
+    Field logDirField;
+    try {
+      logDirField = TaskLog.class.getDeclaredField("LOG_DIR");
+      logDirField.setAccessible(true);
+
+      Field modifiersField = Field.class.getDeclaredField("modifiers");
+      modifiersField.setAccessible(true);
+      modifiersField.setInt(logDirField, logDirField.getModifiers() & ~Modifier.FINAL);
+
+      logDirField.set(null, new File(hadoopLogDir, "userlogs"));
+    } catch (SecurityException e) {
+      throw new RuntimeException(e);
+    } catch (NoSuchFieldException e) {
+      // TODO Auto-generated catch block
+      throw new RuntimeException(e);
+    } catch (IllegalArgumentException e) {
+      throw new RuntimeException(e);
+    } catch (IllegalAccessException e) {
+      throw new RuntimeException(e);
+    }
+  }
+
+  /**
+   * Starts a <code>MiniMRCluster</code>. Call {@link #setFileSystemURI(String)} to use a different
+   * filesystem.
    * @param servers  The number of <code>TaskTracker</code>'s to start.
    * @throws IOException When starting the cluster fails.
    */
-  public void startMiniMapReduceCluster(final int servers) throws IOException {
-    LOG.info("Starting mini mapreduce cluster...");
-    if (dataTestDir == null) {
-      setupDataTestDir();
+  private void startMiniMapReduceCluster(final int servers) throws IOException {
+    if (mrCluster != null) {
+      throw new IllegalStateException("MiniMRCluster is already running");
     }
+    LOG.info("Starting mini mapreduce cluster...");
+    setupClusterTestDir();
+    createDirsAndSetProperties();
 
-    // These are needed for the new and improved Map/Reduce framework
-    conf.set("mapred.output.dir", conf.get("hadoop.tmp.dir"));
-    mrCluster = new MiniMRCluster(0, 0, servers,
-      FileSystem.get(conf).getUri().toString(), 1, null, null, null, new JobConf(conf));
+    forceChangeTaskLogDir();
+
+    // Allow the user to override FS URI for this map-reduce cluster to use.
+    mrCluster = new MiniMRCluster(servers,
+      FS_URI != null ? FS_URI : FileSystem.get(conf).getUri().toString(), 1);
     mrCluster.getJobTrackerRunner().getJobTracker().getConf().set("mapred.local.dir",
-        conf.get("mapred.local.dir")); //Hadoop MiniMR overwrites this while it should not
+      conf.get("mapred.local.dir")); //Hadoop MiniMR overwrites this while it should not
     LOG.info("Mini mapreduce cluster started");
     conf.set("mapred.job.tracker",
         mrCluster.createJobConf().get("mapred.job.tracker"));
@@ -1524,17 +1564,15 @@ public class HBaseTestingUtility {
   }
 
   /**
-   * @return True if we removed the test dir
+   * @return True if we removed the test dirs
    * @throws IOException
    */
-  public boolean cleanupTestDir() throws IOException {
-    if (dataTestDir == null ){
-      return false;
-    } else {
-      boolean ret = deleteDir(getDataTestDir());
-      dataTestDir = null;
-      return ret;
+  boolean cleanupTestDir() throws IOException {
+    if (deleteDir(this.dataTestDir)) {
+      this.dataTestDir = null;
+      return true;
     }
+    return false;
   }
 
   /**
@@ -1542,11 +1580,11 @@ public class HBaseTestingUtility {
    * @return True if we removed the test dir
    * @throws IOException
    */
-  public boolean cleanupTestDir(final String subdir) throws IOException {
-    if (dataTestDir == null){
+  boolean cleanupTestDir(final String subdir) throws IOException {
+    if (this.dataTestDir == null){
       return false;
     }
-    return deleteDir(getDataTestDir(subdir));
+    return deleteDir(new File(this.dataTestDir, subdir));
   }
 
   /**
@@ -1554,10 +1592,15 @@ public class HBaseTestingUtility {
    * @return True if we deleted it.
    * @throws IOException
    */
-  public boolean deleteDir(final Path dir) throws IOException {
-    FileSystem fs = getTestFileSystem();
-    if (fs.exists(dir)) {
-      return fs.delete(getDataTestDir(), true);
+  boolean deleteDir(final File dir) throws IOException {
+    if (dir != null && dir.exists()) {
+      // Need to use deleteDirectory because File.delete required dir is empty.
+      if (!FSUtils.deleteDirectory(FileSystem.getLocal(this.conf),
+          new Path(dir.getAbsolutePath()))) {
+        LOG.warn("Failed delete of " + dir.toString());
+      } else {
+        return true;
+      }
     }
     return false;
   }
@@ -2037,4 +2080,7 @@ public class HBaseTestingUtility {
     return region;
   }
 
+  public void setFileSystemURI(String fsURI) {
+    FS_URI = fsURI;
+  }
 }

src/test/java/org/apache/hadoop/hbase/TestFullLogReconstruction.java

@@ -58,6 +58,7 @@ public class TestFullLogReconstruction {
     // faster failover with cluster.shutdown();fs.close() idiom
     c.setInt("ipc.client.connect.max.retries", 1);
     c.setInt("dfs.client.block.recovery.retries", 1);
+    c.setInt(HConstants.ZOOKEEPER_SESSION_TIMEOUT, 1000);
     TEST_UTIL.startMiniCluster(2);
   }
 
@@ -129,4 +130,3 @@ public class TestFullLogReconstruction {
   public org.apache.hadoop.hbase.ResourceCheckerJUnitRule cu =
     new org.apache.hadoop.hbase.ResourceCheckerJUnitRule();
 }
-

src/test/java/org/apache/hadoop/hbase/TestHBaseTestingUtility.java

@@ -24,9 +24,6 @@ import static org.junit.Assert.assertEquals;
 import static org.junit.Assert.assertFalse;
 import static org.junit.Assert.assertTrue;
 
-import java.io.File;
-import java.io.IOException;
-
 import org.apache.commons.logging.Log;
 import org.apache.commons.logging.LogFactory;
 import org.apache.hadoop.fs.FileSystem;
@@ -38,10 +35,6 @@ import org.apache.hadoop.hbase.client.Result;
 import org.apache.hadoop.hbase.util.Bytes;
 import org.apache.hadoop.hbase.zookeeper.MiniZooKeeperCluster;
 import org.apache.hadoop.hdfs.MiniDFSCluster;
-import org.junit.After;
-import org.junit.AfterClass;
-import org.junit.Before;
-import org.junit.BeforeClass;
 import org.junit.Test;
 import org.junit.experimental.categories.Category;
 
@@ -194,7 +187,7 @@ public class TestHBaseTestingUtility {
 
   @Test public void testMiniDFSCluster() throws Exception {
     HBaseTestingUtility hbt = new HBaseTestingUtility();
-    MiniDFSCluster cluster = hbt.startMiniDFSCluster(1);
+    MiniDFSCluster cluster = hbt.startMiniDFSCluster(null);
     FileSystem dfs = cluster.getFileSystem();
     Path dir = new Path("dir");
     Path qualifiedDir = dfs.makeQualified(dir);
@@ -213,12 +206,11 @@ public class TestHBaseTestingUtility {
 
     assertFalse(fs.exists(testdir));
 
-    hbt.startMiniDFSCluster(1);
+    hbt.startMiniDFSCluster(null);
     assertTrue(fs.exists(testdir));
 
     hbt.shutdownMiniCluster();
     assertFalse(fs.exists(testdir));
-
   }
 
   @Test public void testTestDir() throws Exception {

src/test/java/org/apache/hadoop/hbase/TestZooKeeper.java

@@ -68,8 +68,10 @@ public class TestZooKeeper {
   @BeforeClass
   public static void setUpBeforeClass() throws Exception {
     // Test we can first start the ZK cluster by itself
+    Configuration conf = TEST_UTIL.getConfiguration();
     TEST_UTIL.startMiniZKCluster();
-    TEST_UTIL.getConfiguration().setBoolean("dfs.support.append", true);
+    conf.setBoolean("dfs.support.append", true);
+    conf.setInt(HConstants.ZOOKEEPER_SESSION_TIMEOUT, 1000);
     TEST_UTIL.startMiniCluster(2);
   }
 

src/test/java/org/apache/hadoop/hbase/coprocessor/TestClassLoading.java

@@ -176,6 +176,8 @@ public class TestClassLoading {
     String classpath =
         currentDir + Path.SEPARATOR + "target"+ Path.SEPARATOR + "classes" +
         System.getProperty("path.separator") +
+        // Note that the below trick only works if mvn is running the test;
+        // doesn't work in eclipse for example.
         System.getProperty("surefire.test.class.path");
     options.add(classpath);
     LOG.debug("Setting classpath to: "+classpath);

src/test/java/org/apache/hadoop/hbase/coprocessor/TestCoprocessorEndpoint.java

@@ -67,7 +67,6 @@ public class TestCoprocessorEndpoint {
   private static byte[][] ROWS = makeN(ROW, ROWSIZE);
 
   private static HBaseTestingUtility util = new HBaseTestingUtility();
-  private static MiniHBaseCluster cluster = null;
 
   @BeforeClass
   public static void setupBeforeClass() throws Exception {
@@ -76,10 +75,7 @@ public class TestCoprocessorEndpoint {
     conf.setStrings(CoprocessorHost.REGION_COPROCESSOR_CONF_KEY,
         "org.apache.hadoop.hbase.coprocessor.ColumnAggregationEndpoint",
         "org.apache.hadoop.hbase.coprocessor.GenericEndpoint");
-
     util.startMiniCluster(2);
-    cluster = util.getMiniHBaseCluster();
-
     HTable table = util.createTable(TEST_TABLE, TEST_FAMILY);
     util.createMultiRegions(util.getConfiguration(), table, TEST_FAMILY,
                             new byte[][] { HConstants.EMPTY_BYTE_ARRAY,

src/test/java/org/apache/hadoop/hbase/io/hfile/TestFixedFileTrailer.java

@@ -130,8 +130,7 @@ public class TestFixedFileTrailer {
     }
 
     // Now check what happens if the trailer is corrupted.
-    Path trailerPath = new Path(util.getDataTestDir(), "trailer_"
-        + version);
+    Path trailerPath = new Path(util.getDataTestDir(), "trailer_" + version);
 
     {
       for (byte invalidVersion : new byte[] { HFile.MIN_FORMAT_VERSION - 1,

src/test/java/org/apache/hadoop/hbase/io/hfile/TestHFile.java

@@ -52,7 +52,9 @@ import org.junit.experimental.categories.Category;
 public class TestHFile extends HBaseTestCase {
   static final Log LOG = LogFactory.getLog(TestHFile.class);
 
-  private String ROOT_DIR;
+  private static final HBaseTestingUtility TEST_UTIL = new HBaseTestingUtility();
+  private static String ROOT_DIR =
+    TEST_UTIL.getDataTestDir("TestHFile").toString();
   private final int minBlockSize = 512;
   private static String localFormatter = "%010d";
   private static CacheConfig cacheConf = null;
@@ -61,7 +63,6 @@ public class TestHFile extends HBaseTestCase {
   @Override
   public void setUp() throws Exception {
     startingMetrics = SchemaMetrics.getMetricsSnapshot();
-    ROOT_DIR = this.getUnitTestdir("TestHFile").toString();
     super.setUp();
   }
 

src/test/java/org/apache/hadoop/hbase/io/hfile/TestHFileDataBlockEncoder.java

@@ -78,14 +78,6 @@ public class TestHFileDataBlockEncoder {
     SchemaMetrics.configureGlobally(conf);
   }
 
-  /**
-   * Cleanup after JUnit test.
-   */
-  @After
-  public void tearDown() throws IOException {
-    TEST_UTIL.cleanupTestDir();
-  }
-
   /**
    * Test putting and taking out blocks into cache with different
    * encoding options.

src/test/java/org/apache/hadoop/hbase/io/hfile/TestHFileSeek.java

@@ -66,6 +66,8 @@ public class TestHFileSeek extends TestCase {
   private RandomDistribution.DiscreteRNG keyLenGen;
   private KVGenerator kvGen;
 
+  private static final HBaseTestingUtility TEST_UTIL = new HBaseTestingUtility();
+
   @Override
   public void setUp() throws IOException {
     if (options == null) {

src/test/java/org/apache/hadoop/hbase/io/hfile/TestScannerSelectionUsingTTL.java

@@ -38,7 +38,6 @@ import org.apache.hadoop.hbase.client.Scan;
 import org.apache.hadoop.hbase.io.hfile.BlockType.BlockCategory;
 import org.apache.hadoop.hbase.regionserver.HRegion;
 import org.apache.hadoop.hbase.regionserver.InternalScanner;
-import org.apache.hadoop.hbase.regionserver.StoreFile.BloomType;
 import org.apache.hadoop.hbase.regionserver.metrics.SchemaMetrics;
 import org.apache.hadoop.hbase.regionserver.metrics.SchemaMetrics.BlockMetricType;
 import org.apache.hadoop.hbase.util.Bytes;
@@ -66,7 +65,7 @@ public class TestScannerSelectionUsingTTL {
   private static String FAMILY = "myCF";
   private static byte[] FAMILY_BYTES = Bytes.toBytes(FAMILY);
 
-  private static final int TTL_SECONDS = 2;
+  private static final int TTL_SECONDS = 10;
   private static final int TTL_MS = TTL_SECONDS * 1000;
 
   private static final int NUM_EXPIRED_FILES = 2;
@@ -106,7 +105,7 @@ public class TestScannerSelectionUsingTTL {
     htd.addFamily(hcd);
     HRegionInfo info = new HRegionInfo(Bytes.toBytes(TABLE));
     HRegion region =
-        HRegion.createHRegion(info, TEST_UTIL.getClusterTestDir(),
+        HRegion.createHRegion(info, TEST_UTIL.getDataTestDir(info.getEncodedName()),
             TEST_UTIL.getConfiguration(), htd);
 
     for (int iFile = 0; iFile < totalNumFiles; ++iFile) {

src/test/java/org/apache/hadoop/hbase/mapreduce/TestHFileOutputFormat.java

@@ -77,7 +77,6 @@ import org.apache.hadoop.mapreduce.RecordWriter;
 import org.apache.hadoop.mapreduce.TaskAttemptContext;
 import org.apache.hadoop.mapreduce.TaskAttemptID;
 import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;
-import org.junit.Before;
 import org.junit.Test;
 import org.junit.experimental.categories.Category;
 import org.mockito.Mockito;
@@ -159,12 +158,6 @@ public class TestHFileOutputFormat  {
     }
   }
 
-  @Before
-  public void cleanupDir() throws IOException {
-    util.cleanupTestDir();
-  }
-
-
   private void setupRandomGeneratorMapper(Job job) {
     job.setInputFormatClass(NMapInputFormat.class);
     job.setMapperClass(RandomKVGeneratingMapper.class);
@@ -370,16 +363,19 @@ public class TestHFileOutputFormat  {
 
   @Test
   public void testMRIncrementalLoad() throws Exception {
+    LOG.info("\nStarting test testMRIncrementalLoad\n");
     doIncrementalLoadTest(false);
   }
 
   @Test
   public void testMRIncrementalLoadWithSplit() throws Exception {
+    LOG.info("\nStarting test testMRIncrementalLoadWithSplit\n");
     doIncrementalLoadTest(true);
   }
 
   private void doIncrementalLoadTest(
       boolean shouldChangeRegions) throws Exception {
+    util = new HBaseTestingUtility();
     Configuration conf = util.getConfiguration();
     Path testDir = util.getDataTestDir("testLocalMRIncrementalLoad");
     byte[][] startKeys = generateRandomStartKeys(5);
@@ -442,9 +438,7 @@ public class TestHFileOutputFormat  {
           expectedRows, util.countRows(table));
       Scan scan = new Scan();
       ResultScanner results = table.getScanner(scan);
-      int count = 0;
       for (Result res : results) {
-        count++;
         assertEquals(FAMILIES.length, res.raw().length);
         KeyValue first = res.raw()[0];
         for (KeyValue kv : res.raw()) {

src/test/java/org/apache/hadoop/hbase/mapreduce/TestTimeRangeMapRed.java

@@ -165,7 +165,7 @@ public class TestTimeRangeMapRed {
 
   private void runTestOnTable()
   throws IOException, InterruptedException, ClassNotFoundException {
-    UTIL.startMiniMapReduceCluster(1);
+    UTIL.startMiniMapReduceCluster();
     Job job = null;
     try {
       job = new Job(UTIL.getConfiguration(), "test123");

src/test/java/org/apache/hadoop/hbase/master/TestSplitLogManager.java

@@ -69,6 +69,12 @@ public class TestSplitLogManager {
   private final static HBaseTestingUtility TEST_UTIL =
     new HBaseTestingUtility();
 
+  /**
+   * Additional amount of time we wait for events to happen. Added where unit
+   * test failures have been observed.
+   */
+  private static final int EXTRA_TOLERANCE_MS = 200;
+
   static Stoppable stopper = new Stoppable() {
     @Override
     public void stop(String why) {
@@ -94,7 +100,8 @@ public class TestSplitLogManager {
   public void setup() throws Exception {
     TEST_UTIL.startMiniZKCluster();
     conf = TEST_UTIL.getConfiguration();
-    zkw = new ZooKeeperWatcher(conf, "split-log-manager-tests", null);
+    // Use a different ZK wrapper instance for each tests.
+    zkw = new ZooKeeperWatcher(conf, "split-log-manager-tests" + UUID.randomUUID().toString(), null);
     ZKUtil.deleteChildrenRecursively(zkw, zkw.baseZNode);
     ZKUtil.createAndFailSilent(zkw, zkw.baseZNode);
     assertTrue(ZKUtil.checkExists(zkw, zkw.baseZNode) != -1);
@@ -211,7 +218,7 @@ public class TestSplitLogManager {
     assertTrue((task.last_update <= curt) &&
         (task.last_update > (curt - 1000)));
     LOG.info("waiting for manager to resubmit the orphan task");
-    waitForCounter(tot_mgr_resubmit, 0, 1, to + 100);
+    waitForCounter(tot_mgr_resubmit, 0, 1, to + 300);
     assertTrue(task.isUnassigned());
     waitForCounter(tot_mgr_rescan, 0, 1, to + 100);
   }
@@ -265,7 +272,7 @@ public class TestSplitLogManager {
 
     ZKUtil.setData(zkw, tasknode, TaskState.TASK_OWNED.get("worker1"));
     waitForCounter(tot_mgr_heartbeat, 0, 1, 1000);
-    waitForCounter(tot_mgr_resubmit, 0, 1, to + 100);
+    waitForCounter(tot_mgr_resubmit, 0, 1, to + EXTRA_TOLERANCE_MS);
     int version1 = ZKUtil.checkExists(zkw, tasknode);
     assertTrue(version1 > version);
     ZKUtil.setData(zkw, tasknode, TaskState.TASK_OWNED.get("worker2"));
@@ -275,8 +282,8 @@ public class TestSplitLogManager {
     assertTrue(version2 > version1);
     ZKUtil.setData(zkw, tasknode, TaskState.TASK_OWNED.get("worker3"));
     waitForCounter(tot_mgr_heartbeat, 1, 2, 1000);
-    waitForCounter(tot_mgr_resubmit_threshold_reached, 0, 1, to + 100);
-    Thread.sleep(to + 100);
+    waitForCounter(tot_mgr_resubmit_threshold_reached, 0, 1, to + EXTRA_TOLERANCE_MS);
+    Thread.sleep(to + EXTRA_TOLERANCE_MS);
     assertEquals(2L, tot_mgr_resubmit.get());
   }
 

src/test/java/org/apache/hadoop/hbase/regionserver/TestStoreFile.java

@@ -35,6 +35,7 @@ import org.apache.hadoop.conf.Configuration;
 import org.apache.hadoop.fs.FileSystem;
 import org.apache.hadoop.fs.Path;
 import org.apache.hadoop.hbase.HBaseTestCase;
+import org.apache.hadoop.hbase.HBaseTestingUtility;
 import org.apache.hadoop.hbase.KeyValue;
 import org.apache.hadoop.hbase.SmallTests;
 import org.apache.hadoop.hbase.client.Scan;
@@ -66,8 +67,10 @@ import com.google.common.collect.Lists;
 @Category(SmallTests.class)
 public class TestStoreFile extends HBaseTestCase {
   static final Log LOG = LogFactory.getLog(TestStoreFile.class);
-  private CacheConfig cacheConf =  new CacheConfig(conf);
-  private String ROOT_DIR;
+  private static final HBaseTestingUtility TEST_UTIL = new HBaseTestingUtility();
+  private CacheConfig cacheConf =  new CacheConfig(TEST_UTIL.getConfiguration());
+  private static String ROOT_DIR =
+    TEST_UTIL.getDataTestDir("TestStoreFile").toString();
   private Map<String, Long> startingMetrics;
 
   private static final ChecksumType CKTYPE = ChecksumType.CRC32;
@@ -77,7 +80,6 @@ public class TestStoreFile extends HBaseTestCase {
   public void setUp() throws Exception {
     super.setUp();
     startingMetrics = SchemaMetrics.getMetricsSnapshot();
-    ROOT_DIR = new Path(this.testDir, "TestStoreFile").toString();
   }
 
   @Override

src/test/java/org/apache/hadoop/hbase/regionserver/wal/TestHLog.java

@@ -19,10 +19,7 @@
  */
 package org.apache.hadoop.hbase.regionserver.wal;
 
-import static org.junit.Assert.assertEquals;
-import static org.junit.Assert.assertTrue;
-import static org.junit.Assert.assertNotNull;
-import static org.junit.Assert.fail;
+import static org.junit.Assert.*;
 
 import java.io.IOException;
 import java.lang.reflect.Method;
@@ -51,7 +48,6 @@ import org.apache.hadoop.hdfs.DistributedFileSystem;
 import org.apache.hadoop.hdfs.MiniDFSCluster;
 import org.apache.hadoop.hdfs.protocol.FSConstants;
 import org.apache.hadoop.hdfs.server.datanode.DataNode;
-import org.apache.hadoop.hdfs.server.namenode.FSNamesystem;
 import org.apache.hadoop.hdfs.server.namenode.LeaseManager;
 import org.apache.hadoop.io.SequenceFile;
 import org.apache.log4j.Level;
@@ -405,6 +401,7 @@ public class TestHLog  {
         LOG.error("Waiting for cluster to go down");
         Thread.sleep(1000);
       }
+      assertFalse(cluster.isClusterUp());
 
       // Workaround a strange issue with Hadoop's RPC system - if we don't
       // sleep here, the new datanodes will pick up a cached IPC connection to
@@ -412,11 +409,12 @@ public class TestHLog  {
       // the idle time threshold configured in the conf above
       Thread.sleep(2000);
 
-      cluster = new MiniDFSCluster(namenodePort, conf, 5, false, true, true, null, null, null, null);
-      TEST_UTIL.setDFSCluster(cluster);
+      LOG.info("Waiting a few seconds before re-starting HDFS");
+      Thread.sleep(5000);
+      cluster = TEST_UTIL.startMiniDFSClusterForTestHLog(namenodePort);
       cluster.waitActive();
       fs = cluster.getFileSystem();
-      LOG.info("START second instance.");
+      LOG.info("STARTED second instance.");
     }
 
     // set the lease period to be 1 second so that the

src/test/java/org/apache/hadoop/hbase/util/TestMiniClusterLoadSequential.java

@@ -76,6 +76,9 @@ public class TestMiniClusterLoadSequential {
     this.isMultiPut = isMultiPut;
     this.dataBlockEncoding = dataBlockEncoding;
     conf.setInt(HConstants.HREGION_MEMSTORE_FLUSH_SIZE, 1024 * 1024);
+
+    // We don't want any region reassignments by the load balancer during the test.
+    conf.setFloat(HConstants.LOAD_BALANCER_SLOP_KEY, 10.0f);
   }
 
   @Parameters

src/test/java/org/apache/hadoop/hbase/zookeeper/TestZKLeaderManager.java

@@ -93,7 +93,7 @@ public class TestZKLeaderManager {
 
         while (master.get() && !stopped) {
           try {
-            Thread.sleep(200);
+            Thread.sleep(10);
           } catch (InterruptedException ignored) {}
         }
       }
@@ -204,8 +204,8 @@ public class TestZKLeaderManager {
   private MockLeader getCurrentLeader() throws Exception {
     MockLeader currentLeader = null;
     outer:
-    // wait up to 2 secs for initial leader
-    for (int i = 0; i < 20; i++) {
+    // Wait up to 10 secs for initial leader
+    for (int i = 0; i < 1000; i++) {
       for (int j = 0; j < CANDIDATES.length; j++) {
         if (CANDIDATES[j].isMaster()) {
           // should only be one leader
@@ -218,7 +218,7 @@ public class TestZKLeaderManager {
       if (currentLeader != null) {
         break outer;
       }
-      Thread.sleep(100);
+      Thread.sleep(10);
     }
     return currentLeader;
   }

src/test/resources/hbase-site.xml

@@ -142,4 +142,11 @@
     version is X.X.X-SNAPSHOT"
     </description>
   </property>
+   <property>
+     <name>hbase.client.retries.number</name>
+     <value>100</value>
+     <description>
+     Use a lot of retries in unit tests.
+     </description>
+   </property>
 </configuration>