001/*
002 * Licensed to the Apache Software Foundation (ASF) under one
003 * or more contributor license agreements.  See the NOTICE file
004 * distributed with this work for additional information
005 * regarding copyright ownership.  The ASF licenses this file
006 * to you under the Apache License, Version 2.0 (the
007 * "License"); you may not use this file except in compliance
008 * with the License.  You may obtain a copy of the License at
009 *
010 *     http://www.apache.org/licenses/LICENSE-2.0
011 *
012 * Unless required by applicable law or agreed to in writing, software
013 * distributed under the License is distributed on an "AS IS" BASIS,
014 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
015 * See the License for the specific language governing permissions and
016 * limitations under the License.
017 */
018package org.apache.hadoop.hbase.master;
019
020import static org.junit.Assert.assertEquals;
021import static org.junit.Assert.assertNotEquals;
022import static org.junit.Assert.assertTrue;
023
024import java.io.IOException;
025import org.apache.hadoop.conf.Configuration;
026import org.apache.hadoop.hbase.HBaseClassTestRule;
027import org.apache.hadoop.hbase.HBaseTestingUtility;
028import org.apache.hadoop.hbase.HRegionInfo;
029import org.apache.hadoop.hbase.MiniHBaseCluster;
030import org.apache.hadoop.hbase.MiniHBaseCluster.MiniHBaseClusterRegionServer;
031import org.apache.hadoop.hbase.ServerName;
032import org.apache.hadoop.hbase.StartMiniClusterOption;
033import org.apache.hadoop.hbase.Waiter;
034import org.apache.hadoop.hbase.master.assignment.RegionStates;
035import org.apache.hadoop.hbase.testclassification.MediumTests;
036import org.apache.hadoop.hbase.util.Bytes;
037import org.apache.hadoop.hbase.zookeeper.MetaTableLocator;
038import org.apache.hadoop.hbase.zookeeper.ZKUtil;
039import org.apache.hadoop.hbase.zookeeper.ZNodePaths;
040import org.apache.zookeeper.KeeperException;
041import org.junit.AfterClass;
042import org.junit.BeforeClass;
043import org.junit.ClassRule;
044import org.junit.Test;
045import org.junit.experimental.categories.Category;
046import org.slf4j.Logger;
047import org.slf4j.LoggerFactory;
048
049/**
050 * Tests handling of meta-carrying region server failover.
051 */
052@Category(MediumTests.class)
053public class TestMetaShutdownHandler {
054  private static final Logger LOG = LoggerFactory.getLogger(TestMetaShutdownHandler.class);
055  @ClassRule
056  public static final HBaseClassTestRule CLASS_RULE =
057      HBaseClassTestRule.forClass(TestMetaShutdownHandler.class);
058
059  private final static HBaseTestingUtility TEST_UTIL = new HBaseTestingUtility();
060  final static Configuration conf = TEST_UTIL.getConfiguration();
061
062  @BeforeClass
063  public static void setUpBeforeClass() throws Exception {
064    StartMiniClusterOption option = StartMiniClusterOption.builder()
065        .numRegionServers(3).rsClass(MyRegionServer.class).numDataNodes(3).build();
066    TEST_UTIL.startMiniCluster(option);
067  }
068
069  @AfterClass
070  public static void tearDownAfterClass() throws Exception {
071    TEST_UTIL.shutdownMiniCluster();
072  }
073
074  /**
075   * This test will test the expire handling of a meta-carrying
076   * region server.
077   * After HBaseMiniCluster is up, we will delete the ephemeral
078   * node of the meta-carrying region server, which will trigger
079   * the expire of this region server on the master.
080   * On the other hand, we will slow down the abort process on
081   * the region server so that it is still up during the master SSH.
082   * We will check that the master SSH is still successfully done.
083   */
084  @Test
085  public void testExpireMetaRegionServer() throws Exception {
086    MiniHBaseCluster cluster = TEST_UTIL.getHBaseCluster();
087    HMaster master = cluster.getMaster();
088    RegionStates regionStates = master.getAssignmentManager().getRegionStates();
089    ServerName metaServerName = regionStates.getRegionServerOfRegion(
090      HRegionInfo.FIRST_META_REGIONINFO);
091    if (master.getServerName().equals(metaServerName) || metaServerName == null
092        || !metaServerName.equals(cluster.getServerHoldingMeta())) {
093      // Move meta off master
094      metaServerName =
095          cluster.getLiveRegionServerThreads().get(0).getRegionServer().getServerName();
096      master.move(HRegionInfo.FIRST_META_REGIONINFO.getEncodedNameAsBytes(),
097        Bytes.toBytes(metaServerName.getServerName()));
098      TEST_UTIL.waitUntilNoRegionsInTransition(60000);
099      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}