hbase-server/src/test/java/org/apache/hadoop/hbase/regionserver/TestCompactionInDeadRegionServer.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.regionserver;
  19
  20 import static org.junit.Assert.assertEquals;
  21 import static org.junit.Assert.fail;
  22
  23 import java.io.IOException;
  24 import java.util.Arrays;
  25 import java.util.List;
  26 import org.apache.hadoop.conf.Configuration;
  27 import org.apache.hadoop.hbase.HBaseClassTestRule;
  28 import org.apache.hadoop.hbase.HBaseTestingUtility;
  29 import org.apache.hadoop.hbase.HConstants;
  30 import org.apache.hadoop.hbase.TableName;
  31 import org.apache.hadoop.hbase.Waiter.ExplainingPredicate;
  32 import org.apache.hadoop.hbase.YouAreDeadException;
  33 import org.apache.hadoop.hbase.client.Get;
  34 import org.apache.hadoop.hbase.client.Put;
  35 import org.apache.hadoop.hbase.client.Table;
  36 import org.apache.hadoop.hbase.testclassification.LargeTests;
  37 import org.apache.hadoop.hbase.testclassification.RegionServerTests;
  38 import org.apache.hadoop.hbase.util.Bytes;
  39 import org.apache.hadoop.hbase.util.JVMClusterUtil.RegionServerThread;
  40 import org.apache.hadoop.hbase.wal.AsyncFSWALProvider;
  41 import org.apache.hadoop.hbase.wal.FSHLogProvider;
  42 import org.apache.hadoop.hbase.wal.WALFactory;
  43 import org.apache.hadoop.hbase.wal.WALProvider;
  44 import org.apache.hadoop.hbase.zookeeper.ZKWatcher;
  45 import org.apache.hadoop.hbase.zookeeper.ZNodePaths;
  46 import org.junit.After;
  47 import org.junit.Before;
  48 import org.junit.ClassRule;
  49 import org.junit.Test;
  50 import org.junit.experimental.categories.Category;
  51 import org.junit.runner.RunWith;
  52 import org.junit.runners.Parameterized;
  53 import org.junit.runners.Parameterized.Parameter;
  54 import org.junit.runners.Parameterized.Parameters;
  55 import org.slf4j.Logger;
  56 import org.slf4j.LoggerFactory;
  57
  58 /**
  59  * This testcase is used to ensure that the compaction marker will fail a compaction if the RS is
  60  * already dead. It can not eliminate FNFE when scanning but it does reduce the possibility a lot.
  61  */
  62 @RunWith(Parameterized.class)
  63 @Category({ RegionServerTests.class, LargeTests.class })
  64 public class TestCompactionInDeadRegionServer {
  65
  66   @ClassRule
  67   public static final HBaseClassTestRule CLASS_RULE =
  68       HBaseClassTestRule.forClass(TestCompactionInDeadRegionServer.class);
  69
  70   private static final Logger LOG = LoggerFactory.getLogger(TestCompactionInDeadRegionServer.class);
  71
  72   private static final HBaseTestingUtility UTIL = new HBaseTestingUtility();
  73
  74   private static final TableName TABLE_NAME = TableName.valueOf("test");
  75
  76   private static final byte[] CF = Bytes.toBytes("cf");
  77
  78   private static final byte[] CQ = Bytes.toBytes("cq");
  79
  80   public static final class IgnoreYouAreDeadRS extends HRegionServer {
  81
  82     public IgnoreYouAreDeadRS(Configuration conf) throws IOException, InterruptedException {
  83       super(conf);
  84     }
  85
  86     @Override
  87     protected void tryRegionServerReport(long reportStartTime, long reportEndTime)
  88         throws IOException {
  89       try {
  90         super.tryRegionServerReport(reportStartTime, reportEndTime);
  91       } catch (YouAreDeadException e) {
  92         // ignore, do not abort
  93       }
  94     }
  95   }
  96
  97   @Parameter
  98   public Class<? extends WALProvider> walProvider;
  99
 100   @Parameters(name = "{index}: wal={0}")
 101   public static List<Object[]> params() {
 102     return Arrays.asList(new Object[] { FSHLogProvider.class },
 103         new Object[] { AsyncFSWALProvider.class });
 104   }
 105
 106   @Before
 107   public void setUp() throws Exception {
 108     UTIL.getConfiguration().setClass(WALFactory.WAL_PROVIDER, walProvider, WALProvider.class);
 109     UTIL.getConfiguration().setInt(HConstants.ZK_SESSION_TIMEOUT, 2000);
 110     UTIL.getConfiguration().setClass(HConstants.REGION_SERVER_IMPL, IgnoreYouAreDeadRS.class,
 111       HRegionServer.class);
 112     UTIL.startMiniCluster(2);
 113     Table table = UTIL.createTable(TABLE_NAME, CF);
 114     for (int i = 0; i < 10; i++) {
 115       table.put(new Put(Bytes.toBytes(i)).addColumn(CF, CQ, Bytes.toBytes(i)));
 116     }
 117     UTIL.getAdmin().flush(TABLE_NAME);
 118     for (int i = 10; i < 20; i++) {
 119       table.put(new Put(Bytes.toBytes(i)).addColumn(CF, CQ, Bytes.toBytes(i)));
 120     }
 121     UTIL.getAdmin().flush(TABLE_NAME);
 122   }
 123
 124   @After
 125   public void tearDown() throws Exception {
 126     UTIL.shutdownMiniCluster();
 127   }
 128
 129   @Test
 130   public void test() throws Exception {
 131     HRegionServer regionSvr = UTIL.getRSForFirstRegionInTable(TABLE_NAME);
 132     HRegion region = regionSvr.getRegions(TABLE_NAME).get(0);
 133     String regName = region.getRegionInfo().getEncodedName();
 134     List<HRegion> metaRegs = regionSvr.getRegions(TableName.META_TABLE_NAME);
 135     if (metaRegs != null && !metaRegs.isEmpty()) {
 136       LOG.info("meta is on the same server: " + regionSvr);
 137       // when region is on same server as hbase:meta, reassigning meta would abort the server
 138       // since WAL is broken.
 139       // so the region is moved to a different server
 140       HRegionServer otherRs = UTIL.getOtherRegionServer(regionSvr);
 141       UTIL.moveRegionAndWait(region.getRegionInfo(), otherRs.getServerName());
 142       LOG.info("Moved region: " + regName + " to " + otherRs.getServerName());
 143     }
 144     HRegionServer rsToSuspend = UTIL.getRSForFirstRegionInTable(TABLE_NAME);
 145     region = rsToSuspend.getRegions(TABLE_NAME).get(0);
 146
 147     ZKWatcher watcher = UTIL.getZooKeeperWatcher();
 148     watcher.getRecoverableZooKeeper().delete(
 149       ZNodePaths.joinZNode(watcher.getZNodePaths().rsZNode, rsToSuspend.getServerName().toString()),
 150       -1);
 151     LOG.info("suspending " + rsToSuspend);
 152     UTIL.waitFor(60000, 1000, new ExplainingPredicate<Exception>() {
 153
 154       @Override
 155       public boolean evaluate() throws Exception {
 156         for (RegionServerThread thread : UTIL.getHBaseCluster().getRegionServerThreads()) {
 157           HRegionServer rs = thread.getRegionServer();
 158           if (rs != rsToSuspend) {
 159             return !rs.getRegions(TABLE_NAME).isEmpty();
 160           }
 161         }
 162         return false;
 163       }
 164
 165       @Override
 166       public String explainFailure() throws Exception {
 167         return "The region for " + TABLE_NAME + " is still on " + rsToSuspend.getServerName();
 168       }
 169     });
 170     try {
 171       region.compact(true);
 172       fail("Should fail as our wal file has already been closed, " +
 173           "and walDir has also been renamed");
 174     } catch (Exception e) {
 175       LOG.debug("expected exception: ", e);
 176     }
 177     Table table = UTIL.getConnection().getTable(TABLE_NAME);
 178     // should not hit FNFE
 179     for (int i = 0; i < 20; i++) {
 180       assertEquals(i, Bytes.toInt(table.get(new Get(Bytes.toBytes(i))).getValue(CF, CQ)));
 181     }
 182   }
 183 }