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.assignment;
019
020import java.io.IOException;
021import java.util.List;
022import java.util.concurrent.CountDownLatch;
023import java.util.concurrent.Future;
024import org.apache.hadoop.conf.Configuration;
025import org.apache.hadoop.hbase.HBaseClassTestRule;
026import org.apache.hadoop.hbase.HBaseTestingUtility;
027import org.apache.hadoop.hbase.HConstants;
028import org.apache.hadoop.hbase.ServerName;
029import org.apache.hadoop.hbase.TableName;
030import org.apache.hadoop.hbase.client.RegionInfo;
031import org.apache.hadoop.hbase.master.HMaster;
032import org.apache.hadoop.hbase.master.MasterServices;
033import org.apache.hadoop.hbase.master.RegionPlan;
034import org.apache.hadoop.hbase.master.procedure.RSProcedureDispatcher;
035import org.apache.hadoop.hbase.master.procedure.ServerCrashProcedure;
036import org.apache.hadoop.hbase.master.region.MasterRegion;
037import org.apache.hadoop.hbase.procedure2.ProcedureExecutor;
038import org.apache.hadoop.hbase.testclassification.LargeTests;
039import org.apache.hadoop.hbase.testclassification.MasterTests;
040import org.apache.hadoop.hbase.util.Bytes;
041import org.apache.zookeeper.KeeperException;
042import org.junit.AfterClass;
043import org.junit.BeforeClass;
044import org.junit.ClassRule;
045import org.junit.Test;
046import org.junit.experimental.categories.Category;
047
048/**
049 * Testcase for HBASE-23594.
050 */
051@Category({ MasterTests.class, LargeTests.class })
052public class TestRaceBetweenSCPAndTRSP {
053
054  @ClassRule
055  public static final HBaseClassTestRule CLASS_RULE =
056    HBaseClassTestRule.forClass(TestRaceBetweenSCPAndTRSP.class);
057
058  private static final HBaseTestingUtility UTIL = new HBaseTestingUtility();
059
060  private static TableName NAME = TableName.valueOf("Race");
061
062  private static byte[] CF = Bytes.toBytes("cf");
063
064  private static CountDownLatch ARRIVE_REGION_OPENING;
065
066  private static CountDownLatch RESUME_REGION_OPENING;
067
068  private static CountDownLatch ARRIVE_GET_REGIONS_ON_SERVER;
069
070  private static CountDownLatch RESUME_GET_REGIONS_ON_SERVER;
071
072  private static final class AssignmentManagerForTest extends AssignmentManager {
073
074    public AssignmentManagerForTest(MasterServices master, MasterRegion masterRegion) {
075      super(master, masterRegion);
076    }
077
078    @Override
079    void regionOpening(RegionStateNode regionNode) throws IOException {
080      super.regionOpening(regionNode);
081      if (regionNode.getRegionInfo().getTable().equals(NAME) && ARRIVE_REGION_OPENING != null) {
082        ARRIVE_REGION_OPENING.countDown();
083        ARRIVE_REGION_OPENING = null;
084        try {
085          RESUME_REGION_OPENING.await();
086        } catch (InterruptedException e) {
087        }
088      }
089    }
090
091    @Override
092    public List<RegionInfo> getRegionsOnServer(ServerName serverName) {
093      List<RegionInfo> regions = super.getRegionsOnServer(serverName);
094      if (ARRIVE_GET_REGIONS_ON_SERVER != null) {
095        ARRIVE_GET_REGIONS_ON_SERVER.countDown();
096        ARRIVE_GET_REGIONS_ON_SERVER = null;
097        try {
098          RESUME_GET_REGIONS_ON_SERVER.await();
099        } catch (InterruptedException e) {
100        }
101      }
102      return regions;
103    }
104  }
105
106  public static final class HMasterForTest extends HMaster {
107
108    public HMasterForTest(Configuration conf) throws IOException, KeeperException {
109      super(conf);
110    }
111
112    @Override
113    protected AssignmentManager createAssignmentManager(MasterServices master,
114      MasterRegion masterRegion) {
115      return new AssignmentManagerForTest(master, masterRegion);
116    }
117  }
118
119  @BeforeClass
120  public static void setUp() throws Exception {
121    UTIL.getConfiguration().setClass(HConstants.MASTER_IMPL, HMasterForTest.class, HMaster.class);
122    UTIL.startMiniCluster(2);
123    UTIL.createTable(NAME, CF);
124    UTIL.waitTableAvailable(NAME);
125    UTIL.getAdmin().balancerSwitch(false, true);
126  }
127
128  @AfterClass
129  public static void tearDown() throws Exception {
130    UTIL.shutdownMiniCluster();
131  }
132
133  @Test
134  public void test() throws Exception {
135    RegionInfo region = UTIL.getMiniHBaseCluster().getRegions(NAME).get(0).getRegionInfo();
136    AssignmentManager am = UTIL.getMiniHBaseCluster().getMaster().getAssignmentManager();
137    ServerName sn = am.getRegionStates().getRegionState(region).getServerName();
138
139    // Assign the CountDownLatches that get nulled in background threads else we NPE checking
140    // the static.
141    ARRIVE_REGION_OPENING = new CountDownLatch(1);
142    CountDownLatch arriveRegionOpening = ARRIVE_REGION_OPENING;
143    RESUME_REGION_OPENING = new CountDownLatch(1);
144    ARRIVE_GET_REGIONS_ON_SERVER = new CountDownLatch(1);
145    CountDownLatch arriveGetRegionsOnServer = ARRIVE_GET_REGIONS_ON_SERVER;
146    RESUME_GET_REGIONS_ON_SERVER = new CountDownLatch(1);
147
148    Future<byte[]> moveFuture = am.moveAsync(new RegionPlan(region, sn, sn));
149    arriveRegionOpening.await();
150
151    // Kill the region server and trigger a SCP
152    UTIL.getMiniHBaseCluster().killRegionServer(sn);
153    // Wait until the SCP reaches the getRegionsOnServer call
154    arriveGetRegionsOnServer.await();
155    RSProcedureDispatcher remoteDispatcher = UTIL.getMiniHBaseCluster().getMaster()
156      .getMasterProcedureExecutor().getEnvironment().getRemoteDispatcher();
157    // this is necessary for making the UT stable, the problem here is that, in
158    // ServerManager.expireServer, we will submit the SCP and then the SCP will be executed in
159    // another thread(the PEWorker), so when we reach the above getRegionsOnServer call in SCP, it
160    // is still possible that the expireServer call has not been finished so the remote dispatcher
161    // still think it can dispatcher the TRSP, in this way we will be in dead lock as the TRSP will
162    // not schedule a new ORP since it relies on SCP to wake it up after everything is OK. This is
163    // not what we want to test in this UT so we need to wait here to prevent this from happening.
164    // See HBASE-27277 for more detailed analysis.
165    UTIL.waitFor(15000, () -> !remoteDispatcher.hasNode(sn));
166
167    // Resume the TRSP, it should be able to finish
168    RESUME_REGION_OPENING.countDown();
169    moveFuture.get();
170
171    ProcedureExecutor<?> procExec =
172      UTIL.getMiniHBaseCluster().getMaster().getMasterProcedureExecutor();
173    long scpProcId =
174      procExec.getProcedures().stream().filter(p -> p instanceof ServerCrashProcedure)
175        .map(p -> (ServerCrashProcedure) p).findAny().get().getProcId();
176    // Resume the SCP and make sure it can finish too
177    RESUME_GET_REGIONS_ON_SERVER.countDown();
178    UTIL.waitFor(60000, () -> procExec.isFinished(scpProcId));
179  }
180}