hbase-server/src/test/java/org/apache/hadoop/hbase/master/TestMetaShutdownHandler.java

   1 /*
   2  * Licensed to the Apache Software Foundation (ASF) under one
   3  * or more contributor license agreements.  See the NOTICE file
   4  * distributed with this work for additional information
   5  * regarding copyright ownership.  The ASF licenses this file
   6  * to you under the Apache License, Version 2.0 (the
   7  * "License"); you may not use this file except in compliance
   8  * with the License.  You may obtain a copy of the License at
   9  *
  10  *     http://www.apache.org/licenses/LICENSE-2.0
  11  *
  12  * Unless required by applicable law or agreed to in writing, software
  13  * distributed under the License is distributed on an "AS IS" BASIS,
  14  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  15  * See the License for the specific language governing permissions and
  16  * limitations under the License.
  17  */
  18 package org.apache.hadoop.hbase.master;
  19
  20 import static org.junit.Assert.assertEquals;
  21 import static org.junit.Assert.assertNotEquals;
  22 import static org.junit.Assert.assertTrue;
  23
  24 import java.io.IOException;
  25 import org.apache.hadoop.conf.Configuration;
  26 import org.apache.hadoop.hbase.HBaseClassTestRule;
  27 import org.apache.hadoop.hbase.HBaseTestingUtility;
  28 import org.apache.hadoop.hbase.HRegionInfo;
  29 import org.apache.hadoop.hbase.MiniHBaseCluster;
  30 import org.apache.hadoop.hbase.MiniHBaseCluster.MiniHBaseClusterRegionServer;
  31 import org.apache.hadoop.hbase.ServerName;
  32 import org.apache.hadoop.hbase.StartMiniClusterOption;
  33 import org.apache.hadoop.hbase.Waiter;
  34 import org.apache.hadoop.hbase.master.assignment.RegionStates;
  35 import org.apache.hadoop.hbase.testclassification.MediumTests;
  36 import org.apache.hadoop.hbase.util.Bytes;
  37 import org.apache.hadoop.hbase.zookeeper.MetaTableLocator;
  38 import org.apache.hadoop.hbase.zookeeper.ZKUtil;
  39 import org.apache.hadoop.hbase.zookeeper.ZNodePaths;
  40 import org.apache.zookeeper.KeeperException;
  41 import org.junit.AfterClass;
  42 import org.junit.BeforeClass;
  43 import org.junit.ClassRule;
  44 import org.junit.Test;
  45 import org.junit.experimental.categories.Category;
  46 import org.slf4j.Logger;
  47 import org.slf4j.LoggerFactory;
  48
  49 /**
  50  * Tests handling of meta-carrying region server failover.
  51  */
  52 @Category(MediumTests.class)
  53 public class TestMetaShutdownHandler {
  54   private static final Logger LOG = LoggerFactory.getLogger(TestMetaShutdownHandler.class);
  55   @ClassRule
  56   public static final HBaseClassTestRule CLASS_RULE =
  57       HBaseClassTestRule.forClass(TestMetaShutdownHandler.class);
  58
  59   private final static HBaseTestingUtility TEST_UTIL = new HBaseTestingUtility();
  60   final static Configuration conf = TEST_UTIL.getConfiguration();
  61
  62   @BeforeClass
  63   public static void setUpBeforeClass() throws Exception {
  64     StartMiniClusterOption option = StartMiniClusterOption.builder()
  65         .numRegionServers(3).rsClass(MyRegionServer.class).numDataNodes(3).build();
  66     TEST_UTIL.startMiniCluster(option);
  67   }
  68
  69   @AfterClass
  70   public static void tearDownAfterClass() throws Exception {
  71     TEST_UTIL.shutdownMiniCluster();
  72   }
  73
  74   /**
  75    * This test will test the expire handling of a meta-carrying
  76    * region server.
  77    * After HBaseMiniCluster is up, we will delete the ephemeral
  78    * node of the meta-carrying region server, which will trigger
  79    * the expire of this region server on the master.
  80    * On the other hand, we will slow down the abort process on
  81    * the region server so that it is still up during the master SSH.
  82    * We will check that the master SSH is still successfully done.
  83    */
  84   @Test
  85   public void testExpireMetaRegionServer() throws Exception {
  86     MiniHBaseCluster cluster = TEST_UTIL.getHBaseCluster();
  87     HMaster master = cluster.getMaster();
  88     RegionStates regionStates = master.getAssignmentManager().getRegionStates();
  89     ServerName metaServerName = regionStates.getRegionServerOfRegion(
  90       HRegionInfo.FIRST_META_REGIONINFO);
  91     if (master.getServerName().equals(metaServerName) || metaServerName == null
  92         || !metaServerName.equals(cluster.getServerHoldingMeta())) {
  93       // Move meta off master
  94       metaServerName =
  95           cluster.getLiveRegionServerThreads().get(0).getRegionServer().getServerName();
  96       master.move(HRegionInfo.FIRST_META_REGIONINFO.getEncodedNameAsBytes(),
  97         Bytes.toBytes(metaServerName.getServerName()));
  98       TEST_UTIL.waitUntilNoRegionsInTransition(60000);
  99       metaServerName = regionStates.getRegionServerOfRegion(HRegionInfo.FIRST_META_REGIONINFO);
 100     }
 101     RegionState metaState = MetaTableLocator.getMetaRegionState(master.getZooKeeper());
 102     assertEquals("Wrong state for meta!", RegionState.State.OPEN, metaState.getState());
 103     assertNotEquals("Meta is on master!", metaServerName, master.getServerName());
 104
 105     // Delete the ephemeral node of the meta-carrying region server.
 106     // This is trigger the expire of this region server on the master.
 107     String rsEphemeralNodePath =
 108         ZNodePaths.joinZNode(master.getZooKeeper().getZNodePaths().rsZNode,
 109                 metaServerName.toString());
 110     ZKUtil.deleteNode(master.getZooKeeper(), rsEphemeralNodePath);
 111     LOG.info("Deleted the znode for the RegionServer hosting hbase:meta; waiting on SSH");
 112     // Wait for SSH to finish
 113     final ServerManager serverManager = master.getServerManager();
 114     final ServerName priorMetaServerName = metaServerName;
 115     TEST_UTIL.waitFor(120000, 200, new Waiter.Predicate<Exception>() {
 116       @Override
 117       public boolean evaluate() throws Exception {
 118         return !serverManager.isServerOnline(priorMetaServerName)
 119             && !serverManager.areDeadServersInProgress();
 120       }
 121     });
 122     LOG.info("Past wait on RIT");
 123     TEST_UTIL.waitUntilNoRegionsInTransition(60000);
 124     // Now, make sure meta is assigned
 125     assertTrue("Meta should be assigned",
 126       regionStates.isRegionOnline(HRegionInfo.FIRST_META_REGIONINFO));
 127     // Now, make sure meta is registered in zk
 128     metaState = MetaTableLocator.getMetaRegionState(master.getZooKeeper());
 129     assertEquals("Meta should not be in transition", RegionState.State.OPEN,
 130         metaState.getState());
 131     assertEquals("Meta should be assigned", metaState.getServerName(),
 132       regionStates.getRegionServerOfRegion(HRegionInfo.FIRST_META_REGIONINFO));
 133     assertNotEquals("Meta should be assigned on a different server",
 134       metaState.getServerName(), metaServerName);
 135   }
 136
 137   public static class MyRegionServer extends MiniHBaseClusterRegionServer {
 138
 139     public MyRegionServer(Configuration conf) throws IOException, KeeperException,
 140         InterruptedException {
 141       super(conf);
 142     }
 143
 144     @Override
 145     public void abort(String reason, Throwable cause) {
 146       // sleep to slow down the region server abort
 147       try {
 148         Thread.sleep(30*1000);
 149       } catch (InterruptedException e) {
 150         return;
 151       }
 152       super.abort(reason, cause);
 153     }
 154   }
 155 }