View Javadoc

1   /*
2    * Copyright The Apache Software Foundation
3    *
4    * Licensed to the Apache Software Foundation (ASF) under one
5    * or more contributor license agreements.  See the NOTICE file
6    * distributed with this work for additional information
7    * regarding copyright ownership.  The ASF licenses this file
8    * to you under the Apache License, Version 2.0 (the
9    * "License"); you may not use this file except in compliance
10   * with the License.  You may obtain a copy of the License at
11   *
12   *     http://www.apache.org/licenses/LICENSE-2.0
13   *
14   * Unless required by applicable law or agreed to in writing, software
15   * distributed under the License is distributed on an "AS IS" BASIS,
16   * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
17   * See the License for the specific language governing permissions and
18   * limitations under the License.
19   */
20  package org.apache.hadoop.hbase.replication;
21  
22  import static org.junit.Assert.assertArrayEquals;
23  import static org.junit.Assert.assertEquals;
24  import static org.junit.Assert.fail;
25  
26  import java.io.IOException;
27  
28  import org.apache.commons.logging.Log;
29  import org.apache.commons.logging.LogFactory;
30  import org.apache.hadoop.conf.Configuration;
31  import org.apache.hadoop.hbase.*;
32  import org.apache.hadoop.hbase.client.Delete;
33  import org.apache.hadoop.hbase.client.Get;
34  import org.apache.hadoop.hbase.client.HBaseAdmin;
35  import org.apache.hadoop.hbase.client.HTable;
36  import org.apache.hadoop.hbase.client.Put;
37  import org.apache.hadoop.hbase.client.Result;
38  import org.apache.hadoop.hbase.client.replication.ReplicationAdmin;
39  import org.apache.hadoop.hbase.coprocessor.CoprocessorHost;
40  import org.apache.hadoop.hbase.util.Bytes;
41  import org.apache.hadoop.hbase.zookeeper.MiniZooKeeperCluster;
42  import org.apache.hadoop.hbase.zookeeper.ZooKeeperWatcher;
43  import org.junit.BeforeClass;
44  import org.junit.Test;
45  import org.junit.experimental.categories.Category;
46  
47  @Category(LargeTests.class)
48  public class TestMultiSlaveReplication {
49  
50    private static final Log LOG = LogFactory.getLog(TestReplicationBase.class);
51  
52    private static Configuration conf1;
53    private static Configuration conf2;
54    private static Configuration conf3;
55  
56    private static HBaseTestingUtility utility1;
57    private static HBaseTestingUtility utility2;
58    private static HBaseTestingUtility utility3;
59    private static final long SLEEP_TIME = 500;
60    private static final int NB_RETRIES = 100;
61  
62    private static final byte[] tableName = Bytes.toBytes("test");
63    private static final byte[] famName = Bytes.toBytes("f");
64    private static final byte[] row = Bytes.toBytes("row");
65    private static final byte[] row1 = Bytes.toBytes("row1");
66    private static final byte[] row2 = Bytes.toBytes("row2");
67    private static final byte[] row3 = Bytes.toBytes("row3");
68    private static final byte[] noRepfamName = Bytes.toBytes("norep");
69  
70    private static HTableDescriptor table;
71  
72    @BeforeClass
73    public static void setUpBeforeClass() throws Exception {
74      conf1 = HBaseConfiguration.create();
75      conf1.set(HConstants.ZOOKEEPER_ZNODE_PARENT, "/1");
76      // smaller block size and capacity to trigger more operations
77      // and test them
78      conf1.setInt("hbase.regionserver.hlog.blocksize", 1024*20);
79      conf1.setInt("replication.source.size.capacity", 1024);
80      conf1.setLong("replication.source.sleepforretries", 100);
81      conf1.setInt("hbase.regionserver.maxlogs", 10);
82      conf1.setLong("hbase.master.logcleaner.ttl", 10);
83      conf1.setBoolean(HConstants.REPLICATION_ENABLE_KEY, HConstants.REPLICATION_ENABLE_DEFAULT);
84      conf1.setBoolean("dfs.support.append", true);
85      conf1.setLong(HConstants.THREAD_WAKE_FREQUENCY, 100);
86      conf1.setStrings(CoprocessorHost.USER_REGION_COPROCESSOR_CONF_KEY,
87          "org.apache.hadoop.hbase.replication.TestMasterReplication$CoprocessorCounter");
88  
89      utility1 = new HBaseTestingUtility(conf1);
90      utility1.startMiniZKCluster();
91      MiniZooKeeperCluster miniZK = utility1.getZkCluster();
92      new ZooKeeperWatcher(conf1, "cluster1", null, true);
93  
94      conf2 = new Configuration(conf1);
95      conf2.set(HConstants.ZOOKEEPER_ZNODE_PARENT, "/2");
96  
97      conf3 = new Configuration(conf1);
98      conf3.set(HConstants.ZOOKEEPER_ZNODE_PARENT, "/3");
99  
100     utility2 = new HBaseTestingUtility(conf2);
101     utility2.setZkCluster(miniZK);
102     new ZooKeeperWatcher(conf2, "cluster3", null, true);
103 
104     utility3 = new HBaseTestingUtility(conf3);
105     utility3.setZkCluster(miniZK);
106     new ZooKeeperWatcher(conf3, "cluster3", null, true);
107 
108     table = new HTableDescriptor(TableName.valueOf(tableName));
109     HColumnDescriptor fam = new HColumnDescriptor(famName);
110     fam.setScope(HConstants.REPLICATION_SCOPE_GLOBAL);
111     table.addFamily(fam);
112     fam = new HColumnDescriptor(noRepfamName);
113     table.addFamily(fam);
114   }
115 
116   @Test(timeout=300000)
117   public void testMultiSlaveReplication() throws Exception {
118     LOG.info("testCyclicReplication");
119     MiniHBaseCluster master = utility1.startMiniCluster();
120     utility2.startMiniCluster();
121     utility3.startMiniCluster();
122     ReplicationAdmin admin1 = new ReplicationAdmin(conf1);
123 
124     new HBaseAdmin(conf1).createTable(table);
125     new HBaseAdmin(conf2).createTable(table);
126     new HBaseAdmin(conf3).createTable(table);
127     HTable htable1 = new HTable(conf1, tableName);
128     htable1.setWriteBufferSize(1024);
129     HTable htable2 = new HTable(conf2, tableName);
130     htable2.setWriteBufferSize(1024);
131     HTable htable3 = new HTable(conf3, tableName);
132     htable3.setWriteBufferSize(1024);
133     
134     admin1.addPeer("1", utility2.getClusterKey());
135 
136     // put "row" and wait 'til it got around, then delete
137     putAndWait(row, famName, htable1, htable2);
138     deleteAndWait(row, htable1, htable2);
139     // check it wasn't replication to cluster 3
140     checkRow(row,0,htable3);
141 
142     putAndWait(row2, famName, htable1, htable2);
143 
144     // now roll the region server's logs
145     new HBaseAdmin(conf1).rollHLogWriter(master.getRegionServer(0).getServerName().toString());
146     // after the log was rolled put a new row
147     putAndWait(row3, famName, htable1, htable2);
148 
149     admin1.addPeer("2", utility3.getClusterKey());
150 
151     // put a row, check it was replicated to all clusters
152     putAndWait(row1, famName, htable1, htable2, htable3);
153     // delete and verify
154     deleteAndWait(row1, htable1, htable2, htable3);
155 
156     // make sure row2 did not get replicated after
157     // cluster 3 was added
158     checkRow(row2,0,htable3);
159 
160     // row3 will get replicated, because it was in the
161     // latest log
162     checkRow(row3,1,htable3);
163 
164     Put p = new Put(row);
165     p.add(famName, row, row);
166     htable1.put(p);
167     // now roll the logs again
168     new HBaseAdmin(conf1).rollHLogWriter(master.getRegionServer(0)
169         .getServerName().toString());
170 
171     // cleanup "row2", also conveniently use this to wait replication
172     // to finish
173     deleteAndWait(row2, htable1, htable2, htable3);
174     // Even if the log was rolled in the middle of the replication
175     // "row" is still replication.
176     checkRow(row, 1, htable2);
177     // Replication thread of cluster 2 may be sleeping, and since row2 is not there in it, 
178     // we should wait before checking.
179     checkWithWait(row, 1, htable3);
180 
181     // cleanup the rest
182     deleteAndWait(row, htable1, htable2, htable3);
183     deleteAndWait(row3, htable1, htable2, htable3);
184 
185     utility3.shutdownMiniCluster();
186     utility2.shutdownMiniCluster();
187     utility1.shutdownMiniCluster();
188   }
189  
190   private void checkWithWait(byte[] row, int count, HTable table) throws Exception {
191     Get get = new Get(row);
192     for (int i = 0; i < NB_RETRIES; i++) {
193       if (i == NB_RETRIES - 1) {
194         fail("Waited too much time while getting the row.");
195       }
196       boolean rowReplicated = false;
197       Result res = table.get(get);
198       if (res.size() >= 1) {
199         LOG.info("Row is replicated");
200         rowReplicated = true;
201         assertEquals(count, res.size());
202         break;
203       }
204       if (rowReplicated) {
205         break;
206       } else {
207         Thread.sleep(SLEEP_TIME);
208       }
209     }
210   }
211   
212   private void checkRow(byte[] row, int count, HTable... tables) throws IOException {
213     Get get = new Get(row);
214     for (HTable table : tables) {
215       Result res = table.get(get);
216       assertEquals(count, res.size());
217     }
218   }
219 
220   private void deleteAndWait(byte[] row, HTable source, HTable... targets)
221   throws Exception {
222     Delete del = new Delete(row);
223     source.delete(del);
224 
225     Get get = new Get(row);
226     for (int i = 0; i < NB_RETRIES; i++) {
227       if (i==NB_RETRIES-1) {
228         fail("Waited too much time for del replication");
229       }
230       boolean removedFromAll = true;
231       for (HTable target : targets) {
232         Result res = target.get(get);
233         if (res.size() >= 1) {
234           LOG.info("Row not deleted");
235           removedFromAll = false;
236           break;
237         }
238       }
239       if (removedFromAll) {
240         break;
241       } else {
242         Thread.sleep(SLEEP_TIME);        
243       }
244     }
245   }
246 
247   private void putAndWait(byte[] row, byte[] fam, HTable source, HTable... targets)
248   throws Exception {
249     Put put = new Put(row);
250     put.add(fam, row, row);
251     source.put(put);
252 
253     Get get = new Get(row);
254     for (int i = 0; i < NB_RETRIES; i++) {
255       if (i==NB_RETRIES-1) {
256         fail("Waited too much time for put replication");
257       }
258       boolean replicatedToAll = true;
259       for (HTable target : targets) {
260         Result res = target.get(get);
261         if (res.size() == 0) {
262           LOG.info("Row not available");
263           replicatedToAll = false;
264           break;
265         } else {
266           assertArrayEquals(res.value(), row);
267         }
268       }
269       if (replicatedToAll) {
270         break;
271       } else {
272         Thread.sleep(SLEEP_TIME);
273       }
274     }
275   }
276 
277 }
278