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.wal;
19  
20  import static org.junit.Assert.assertTrue;
21  import static org.junit.Assert.fail;
22  
23  import java.io.IOException;
24  
25  import org.apache.commons.logging.Log;
26  import org.apache.commons.logging.LogFactory;
27  import org.apache.commons.logging.impl.Log4JLogger;
28  import org.apache.hadoop.fs.FileSystem;
29  import org.apache.hadoop.hbase.*;
30  import org.apache.hadoop.hbase.client.HBaseAdmin;
31  import org.apache.hadoop.hbase.client.HTable;
32  import org.apache.hadoop.hbase.client.Put;
33  import org.apache.hadoop.hbase.regionserver.HRegion;
34  import org.apache.hadoop.hbase.regionserver.HRegionServer;
35  import org.apache.hadoop.hbase.util.Bytes;
36  import org.apache.hadoop.hbase.util.FSUtils;
37  import org.apache.hadoop.hdfs.DFSClient;
38  import org.apache.hadoop.hdfs.MiniDFSCluster;
39  import org.apache.hadoop.hdfs.server.datanode.DataNode;
40  import org.apache.hadoop.hdfs.server.namenode.LeaseManager;
41  import org.apache.log4j.Level;
42  import org.junit.After;
43  import org.junit.Before;
44  import org.junit.BeforeClass;
45  import org.junit.Test;
46  import org.junit.experimental.categories.Category;
47  
48  /**
49   * Tests for conditions that should trigger RegionServer aborts when
50   * rolling the current HLog fails.
51   */
52  @Category(MediumTests.class)
53  public class TestLogRollAbort {
54    private static final Log LOG = LogFactory.getLog(TestLogRolling.class);
55    private static MiniDFSCluster dfsCluster;
56    private static HBaseAdmin admin;
57    private static MiniHBaseCluster cluster;
58    private final static HBaseTestingUtility TEST_UTIL = new HBaseTestingUtility();
59  
60    // verbose logging on classes that are touched in these tests
61    {
62      ((Log4JLogger)DataNode.LOG).getLogger().setLevel(Level.ALL);
63      ((Log4JLogger)LeaseManager.LOG).getLogger().setLevel(Level.ALL);
64      ((Log4JLogger)LogFactory.getLog("org.apache.hadoop.hdfs.server.namenode.FSNamesystem"))
65          .getLogger().setLevel(Level.ALL);
66      ((Log4JLogger)DFSClient.LOG).getLogger().setLevel(Level.ALL);
67      ((Log4JLogger)HRegionServer.LOG).getLogger().setLevel(Level.ALL);
68      ((Log4JLogger)HRegion.LOG).getLogger().setLevel(Level.ALL);
69      ((Log4JLogger)HLog.LOG).getLogger().setLevel(Level.ALL);
70    }
71  
72    // Need to override this setup so we can edit the config before it gets sent
73    // to the HDFS & HBase cluster startup.
74    @BeforeClass
75    public static void setUpBeforeClass() throws Exception {
76      // Tweak default timeout values down for faster recovery
77      TEST_UTIL.getConfiguration().setInt(
78          "hbase.regionserver.logroll.errors.tolerated", 2);
79      TEST_UTIL.getConfiguration().setInt("ipc.ping.interval", 10 * 1000);
80      TEST_UTIL.getConfiguration().setInt("ipc.socket.timeout", 10 * 1000);
81      TEST_UTIL.getConfiguration().setInt("hbase.rpc.timeout", 10 * 1000);
82  
83      // Increase the amount of time between client retries
84      TEST_UTIL.getConfiguration().setLong("hbase.client.pause", 5 * 1000);
85  
86      // make sure log.hflush() calls syncFs() to open a pipeline
87      TEST_UTIL.getConfiguration().setBoolean("dfs.support.append", true);
88      // lower the namenode & datanode heartbeat so the namenode
89      // quickly detects datanode failures
90      TEST_UTIL.getConfiguration().setInt("heartbeat.recheck.interval", 5000);
91      TEST_UTIL.getConfiguration().setInt("dfs.heartbeat.interval", 1);
92      // the namenode might still try to choose the recently-dead datanode
93      // for a pipeline, so try to a new pipeline multiple times
94      TEST_UTIL.getConfiguration().setInt("dfs.client.block.write.retries", 10);
95      // set periodic sync to 2 min so it doesn't run during test
96      TEST_UTIL.getConfiguration().setInt("hbase.regionserver.optionallogflushinterval",
97          120 * 1000);
98    }
99  
100   @Before
101   public void setUp() throws Exception {
102     TEST_UTIL.startMiniCluster(2);
103 
104     cluster = TEST_UTIL.getHBaseCluster();
105     dfsCluster = TEST_UTIL.getDFSCluster();
106     admin = TEST_UTIL.getHBaseAdmin();
107 
108     // disable region rebalancing (interferes with log watching)
109     cluster.getMaster().balanceSwitch(false);
110   }
111 
112   @After
113   public void tearDown() throws Exception {
114     TEST_UTIL.shutdownMiniCluster();
115   }
116 
117   /**
118    * Tests that RegionServer aborts if we hit an error closing the WAL when
119    * there are unsynced WAL edits.  See HBASE-4282.
120    */
121   @Test
122   public void testRSAbortWithUnflushedEdits() throws Exception {
123     LOG.info("Starting testRSAbortWithUnflushedEdits()");
124 
125     // When the META table can be opened, the region servers are running
126     new HTable(TEST_UTIL.getConfiguration(),
127       HConstants.META_TABLE_NAME).close();
128 
129     // Create the test table and open it
130     String tableName = this.getClass().getSimpleName();
131     HTableDescriptor desc = new HTableDescriptor(tableName);
132     desc.addFamily(new HColumnDescriptor(HConstants.CATALOG_FAMILY));
133     desc.setDeferredLogFlush(true);
134 
135     admin.createTable(desc);
136     HTable table = new HTable(TEST_UTIL.getConfiguration(), tableName);
137 
138     HRegionServer server = TEST_UTIL.getRSForFirstRegionInTable(Bytes.toBytes(tableName));
139     HLog log = server.getWAL();
140 
141     assertTrue("Need HDFS-826 for this test", log.canGetCurReplicas());
142     // don't run this test without append support (HDFS-200 & HDFS-142)
143     assertTrue("Need append support for this test",
144         FSUtils.isAppendSupported(TEST_UTIL.getConfiguration()));
145 
146     Put p = new Put(Bytes.toBytes("row2001"));
147     p.add(HConstants.CATALOG_FAMILY, Bytes.toBytes("col"), Bytes.toBytes(2001));
148     table.put(p);
149 
150     log.sync();
151 
152     p = new Put(Bytes.toBytes("row2002"));
153     p.add(HConstants.CATALOG_FAMILY, Bytes.toBytes("col"), Bytes.toBytes(2002));
154     table.put(p);
155 
156     dfsCluster.restartDataNodes();
157     LOG.info("Restarted datanodes");
158 
159     assertTrue("Should have an outstanding WAL edit", log.hasDeferredEntries());
160     try {
161       log.rollWriter(true);
162       fail("Log roll should have triggered FailedLogCloseException");
163     } catch (FailedLogCloseException flce) {
164       assertTrue("Should have deferred flush log edits outstanding",
165           log.hasDeferredEntries());
166     }
167   }
168 
169   @org.junit.Rule
170   public org.apache.hadoop.hbase.ResourceCheckerJUnitRule cu =
171     new org.apache.hadoop.hbase.ResourceCheckerJUnitRule();
172 }
173