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.client;
019
020import static org.junit.Assert.assertFalse;
021import static org.junit.Assert.assertNull;
022import static org.junit.Assert.assertTrue;
023
024import java.io.IOException;
025import java.util.ArrayList;
026import java.util.Collections;
027import java.util.List;
028import org.apache.hadoop.conf.Configuration;
029import org.apache.hadoop.hbase.HBaseClassTestRule;
030import org.apache.hadoop.hbase.HBaseTestingUtil;
031import org.apache.hadoop.hbase.HConstants;
032import org.apache.hadoop.hbase.StartTestingClusterOption;
033import org.apache.hadoop.hbase.master.HMaster;
034import org.apache.hadoop.hbase.master.MasterServices;
035import org.apache.hadoop.hbase.master.assignment.AssignmentManager;
036import org.apache.hadoop.hbase.master.assignment.RegionStateNode;
037import org.apache.hadoop.hbase.master.assignment.TransitRegionStateProcedure;
038import org.apache.hadoop.hbase.master.procedure.MasterProcedureEnv;
039import org.apache.hadoop.hbase.master.region.MasterRegion;
040import org.apache.hadoop.hbase.procedure2.Procedure;
041import org.apache.hadoop.hbase.procedure2.ProcedureSuspendedException;
042import org.apache.hadoop.hbase.procedure2.ProcedureYieldException;
043import org.apache.hadoop.hbase.testclassification.MediumTests;
044import org.apache.hadoop.hbase.testclassification.MiscTests;
045import org.junit.AfterClass;
046import org.junit.BeforeClass;
047import org.junit.ClassRule;
048import org.junit.Test;
049import org.junit.experimental.categories.Category;
050
051@Category({ MiscTests.class, MediumTests.class })
052public class TestFailedMetaReplicaAssigment {
053
054  @ClassRule
055  public static final HBaseClassTestRule CLASS_RULE =
056    HBaseClassTestRule.forClass(TestFailedMetaReplicaAssigment.class);
057
058  private static final HBaseTestingUtil TEST_UTIL = new HBaseTestingUtil();
059
060  @BeforeClass
061  public static void setUp() throws Exception {
062    // using our rigged master, to force a failed meta replica assignment when start up master
063    // this test can be removed once we remove the HConstants.META_REPLICAS_NUM config.
064    Configuration conf = TEST_UTIL.getConfiguration();
065    conf.setInt(HConstants.META_REPLICAS_NUM, 3);
066    StartTestingClusterOption option =
067      StartTestingClusterOption.builder().numAlwaysStandByMasters(1).numMasters(1)
068        .numRegionServers(1).masterClass(BrokenMetaReplicaMaster.class).build();
069    TEST_UTIL.startMiniCluster(option);
070  }
071
072  @AfterClass
073  public static void tearDown() throws IOException {
074    TEST_UTIL.shutdownMiniCluster();
075  }
076
077  @Test
078  public void testFailedReplicaAssignment() throws InterruptedException {
079    HMaster master = TEST_UTIL.getMiniHBaseCluster().getMaster();
080    // waiting for master to come up
081    TEST_UTIL.waitFor(30000, () -> master.isInitialized());
082
083    AssignmentManager am = master.getAssignmentManager();
084    // showing one of the replicas got assigned
085    RegionInfo metaReplicaHri =
086      RegionReplicaUtil.getRegionInfoForReplica(RegionInfoBuilder.FIRST_META_REGIONINFO, 1);
087    // we use assignAsync so we need to wait a bit
088    TEST_UTIL.waitFor(30000, () -> {
089      RegionStateNode metaReplicaRegionNode =
090        am.getRegionStates().getOrCreateRegionStateNode(metaReplicaHri);
091      return metaReplicaRegionNode.getRegionLocation() != null;
092    });
093    // showing one of the replicas failed to be assigned
094    RegionInfo metaReplicaHri2 =
095      RegionReplicaUtil.getRegionInfoForReplica(RegionInfoBuilder.FIRST_META_REGIONINFO, 2);
096    RegionStateNode metaReplicaRegionNode2 =
097      am.getRegionStates().getOrCreateRegionStateNode(metaReplicaHri2);
098    // wait for several seconds to make sure that it is not assigned
099    for (int i = 0; i < 3; i++) {
100      Thread.sleep(2000);
101      assertNull(metaReplicaRegionNode2.getRegionLocation());
102    }
103
104    // showing master is active and running
105    assertFalse(master.isStopping());
106    assertFalse(master.isStopped());
107    assertTrue(master.isActiveMaster());
108  }
109
110  public static class BrokenTransitRegionStateProcedure extends TransitRegionStateProcedure {
111
112    public BrokenTransitRegionStateProcedure() {
113      super(null, null, null, false, TransitionType.ASSIGN);
114    }
115
116    public BrokenTransitRegionStateProcedure(MasterProcedureEnv env, RegionInfo hri) {
117      super(env, hri, null, false, TransitionType.ASSIGN);
118    }
119
120    @Override
121    protected Procedure[] execute(MasterProcedureEnv env)
122      throws ProcedureSuspendedException, ProcedureYieldException, InterruptedException {
123      throw new ProcedureSuspendedException("Never end procedure!");
124    }
125  }
126
127  public static class BrokenMetaReplicaMaster extends HMaster {
128    public BrokenMetaReplicaMaster(final Configuration conf) throws IOException {
129      super(conf);
130    }
131
132    @Override
133    public AssignmentManager createAssignmentManager(MasterServices master,
134      MasterRegion masterRegion) {
135      return new BrokenMasterMetaAssignmentManager(master, masterRegion);
136    }
137  }
138
139  public static class BrokenMasterMetaAssignmentManager extends AssignmentManager {
140    MasterServices master;
141
142    public BrokenMasterMetaAssignmentManager(final MasterServices master,
143      MasterRegion masterRegion) {
144      super(master, masterRegion);
145      this.master = master;
146    }
147
148    @Override
149    public TransitRegionStateProcedure[] createAssignProcedures(List<RegionInfo> hris) {
150      List<TransitRegionStateProcedure> procs = new ArrayList<>();
151      for (RegionInfo hri : hris) {
152        if (hri.isMetaRegion() && hri.getReplicaId() == 2) {
153          RegionStateNode regionNode = getRegionStates().getOrCreateRegionStateNode(hri);
154          regionNode.lock();
155          try {
156            procs.add(regionNode.setProcedure(new BrokenTransitRegionStateProcedure(
157              master.getMasterProcedureExecutor().getEnvironment(), hri)));
158          } finally {
159            regionNode.unlock();
160          }
161        } else {
162          procs.add(super.createAssignProcedures(Collections.singletonList(hri))[0]);
163        }
164      }
165      return procs.toArray(new TransitRegionStateProcedure[0]);
166    }
167  }
168}