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;
19
20 import java.io.Closeable;
21 import java.io.IOException;
22
23 import org.apache.commons.logging.Log;
24 import org.apache.commons.logging.LogFactory;
25 import org.apache.hadoop.hbase.classification.InterfaceAudience;
26 import org.apache.hadoop.conf.Configurable;
27 import org.apache.hadoop.conf.Configuration;
28 import org.apache.hadoop.hbase.protobuf.generated.AdminProtos.AdminService;
29 import org.apache.hadoop.hbase.protobuf.generated.ClientProtos.ClientService;
30 import org.apache.hadoop.hbase.protobuf.generated.MasterProtos.MasterService;
31 import org.apache.hadoop.hbase.util.Threads;
32
33 /**
34 * This class defines methods that can help with managing HBase clusters
35 * from unit tests and system tests. There are 3 types of cluster deployments:
36 * <ul>
37 * <li><b>MiniHBaseCluster:</b> each server is run in the same JVM in separate threads,
38 * used by unit tests</li>
39 * <li><b>DistributedHBaseCluster:</b> the cluster is pre-deployed, system and integration tests can
40 * interact with the cluster. </li>
41 * <li><b>ProcessBasedLocalHBaseCluster:</b> each server is deployed locally but in separate
42 * JVMs. </li>
43 * </ul>
44 * <p>
45 * HBaseCluster unifies the way tests interact with the cluster, so that the same test can
46 * be run against a mini-cluster during unit test execution, or a distributed cluster having
47 * tens/hundreds of nodes during execution of integration tests.
48 *
49 * <p>
50 * HBaseCluster exposes client-side public interfaces to tests, so that tests does not assume
51 * running in a particular mode. Not all the tests are suitable to be run on an actual cluster,
52 * and some tests will still need to mock stuff and introspect internal state. For those use
53 * cases from unit tests, or if more control is needed, you can use the subclasses directly.
54 * In that sense, this class does not abstract away <strong>every</strong> interface that
55 * MiniHBaseCluster or DistributedHBaseCluster provide.
56 */
57 @InterfaceAudience.Private
58 public abstract class HBaseCluster implements Closeable, Configurable {
59 // Log is being used in DistributedHBaseCluster class, hence keeping it as package scope
60 static final Log LOG = LogFactory.getLog(HBaseCluster.class.getName());
61 protected Configuration conf;
62
63 /** the status of the cluster before we begin */
64 protected ClusterStatus initialClusterStatus;
65
66 /**
67 * Construct an HBaseCluster
68 * @param conf Configuration to be used for cluster
69 */
70 public HBaseCluster(Configuration conf) {
71 setConf(conf);
72 }
73
74 @Override
75 public void setConf(Configuration conf) {
76 this.conf = conf;
77 }
78
79 @Override
80 public Configuration getConf() {
81 return conf;
82 }
83
84 /**
85 * Returns a ClusterStatus for this HBase cluster.
86 * @see #getInitialClusterStatus()
87 */
88 public abstract ClusterStatus getClusterStatus() throws IOException;
89
90 /**
91 * Returns a ClusterStatus for this HBase cluster as observed at the
92 * starting of the HBaseCluster
93 */
94 public ClusterStatus getInitialClusterStatus() throws IOException {
95 return initialClusterStatus;
96 }
97
98 /**
99 * Returns an {@link MasterService.BlockingInterface} to the active master
100 */
101 public abstract MasterService.BlockingInterface getMasterAdminService()
102 throws IOException;
103
104 /**
105 * Returns an AdminProtocol interface to the regionserver
106 */
107 public abstract AdminService.BlockingInterface getAdminProtocol(ServerName serverName)
108 throws IOException;
109
110 /**
111 * Returns a ClientProtocol interface to the regionserver
112 */
113 public abstract ClientService.BlockingInterface getClientProtocol(ServerName serverName)
114 throws IOException;
115
116 /**
117 * Starts a new region server on the given hostname or if this is a mini/local cluster,
118 * starts a region server locally.
119 * @param hostname the hostname to start the regionserver on
120 * @throws IOException if something goes wrong
121 */
122 public abstract void startRegionServer(String hostname, int port) throws IOException;
123
124 /**
125 * Kills the region server process if this is a distributed cluster, otherwise
126 * this causes the region server to exit doing basic clean up only.
127 * @throws IOException if something goes wrong
128 */
129 public abstract void killRegionServer(ServerName serverName) throws IOException;
130
131 /**
132 * Stops the given region server, by attempting a gradual stop.
133 * @return whether the operation finished with success
134 * @throws IOException if something goes wrong
135 */
136 public abstract void stopRegionServer(ServerName serverName) throws IOException;
137
138 /**
139 * Wait for the specified region server to join the cluster
140 * @return whether the operation finished with success
141 * @throws IOException if something goes wrong or timeout occurs
142 */
143 public void waitForRegionServerToStart(String hostname, int port, long timeout)
144 throws IOException {
145 long start = System.currentTimeMillis();
146 while ((System.currentTimeMillis() - start) < timeout) {
147 for (ServerName server : getClusterStatus().getServers()) {
148 if (server.getHostname().equals(hostname) && server.getPort() == port) {
149 return;
150 }
151 }
152 Threads.sleep(100);
153 }
154 throw new IOException("did timeout " + timeout + "ms waiting for region server to start: "
155 + hostname);
156 }
157
158 /**
159 * Wait for the specified region server to stop the thread / process.
160 * @return whether the operation finished with success
161 * @throws IOException if something goes wrong or timeout occurs
162 */
163 public abstract void waitForRegionServerToStop(ServerName serverName, long timeout)
164 throws IOException;
165
166 /**
167 * Starts a new zookeeper node on the given hostname or if this is a mini/local cluster,
168 * silently logs warning message.
169 * @param hostname the hostname to start the regionserver on
170 * @throws IOException if something goes wrong
171 */
172 public abstract void startZkNode(String hostname, int port) throws IOException;
173
174 /**
175 * Kills the zookeeper node process if this is a distributed cluster, otherwise,
176 * this causes master to exit doing basic clean up only.
177 * @throws IOException if something goes wrong
178 */
179 public abstract void killZkNode(ServerName serverName) throws IOException;
180
181 /**
182 * Stops the region zookeeper if this is a distributed cluster, otherwise
183 * silently logs warning message.
184 * @throws IOException if something goes wrong
185 */
186 public abstract void stopZkNode(ServerName serverName) throws IOException;
187
188 /**
189 * Wait for the specified zookeeper node to join the cluster
190 * @return whether the operation finished with success
191 * @throws IOException if something goes wrong or timeout occurs
192 */
193 public abstract void waitForZkNodeToStart(ServerName serverName, long timeout)
194 throws IOException;
195
196 /**
197 * Wait for the specified zookeeper node to stop the thread / process.
198 * @return whether the operation finished with success
199 * @throws IOException if something goes wrong or timeout occurs
200 */
201 public abstract void waitForZkNodeToStop(ServerName serverName, long timeout)
202 throws IOException;
203
204 /**
205 * Starts a new datanode on the given hostname or if this is a mini/local cluster,
206 * silently logs warning message.
207 * @throws IOException if something goes wrong
208 */
209 public abstract void startDataNode(ServerName serverName) throws IOException;
210
211 /**
212 * Kills the datanode process if this is a distributed cluster, otherwise,
213 * this causes master to exit doing basic clean up only.
214 * @throws IOException if something goes wrong
215 */
216 public abstract void killDataNode(ServerName serverName) throws IOException;
217
218 /**
219 * Stops the datanode if this is a distributed cluster, otherwise
220 * silently logs warning message.
221 * @throws IOException if something goes wrong
222 */
223 public abstract void stopDataNode(ServerName serverName) throws IOException;
224
225 /**
226 * Wait for the specified datanode to join the cluster
227 * @return whether the operation finished with success
228 * @throws IOException if something goes wrong or timeout occurs
229 */
230 public abstract void waitForDataNodeToStart(ServerName serverName, long timeout)
231 throws IOException;
232
233 /**
234 * Wait for the specified datanode to stop the thread / process.
235 * @return whether the operation finished with success
236 * @throws IOException if something goes wrong or timeout occurs
237 */
238 public abstract void waitForDataNodeToStop(ServerName serverName, long timeout)
239 throws IOException;
240
241 /**
242 * Starts a new master on the given hostname or if this is a mini/local cluster,
243 * starts a master locally.
244 * @param hostname the hostname to start the master on
245 * @return whether the operation finished with success
246 * @throws IOException if something goes wrong
247 */
248 public abstract void startMaster(String hostname, int port) throws IOException;
249
250 /**
251 * Kills the master process if this is a distributed cluster, otherwise,
252 * this causes master to exit doing basic clean up only.
253 * @throws IOException if something goes wrong
254 */
255 public abstract void killMaster(ServerName serverName) throws IOException;
256
257 /**
258 * Stops the given master, by attempting a gradual stop.
259 * @throws IOException if something goes wrong
260 */
261 public abstract void stopMaster(ServerName serverName) throws IOException;
262
263 /**
264 * Wait for the specified master to stop the thread / process.
265 * @throws IOException if something goes wrong or timeout occurs
266 */
267 public abstract void waitForMasterToStop(ServerName serverName, long timeout)
268 throws IOException;
269
270 /**
271 * Blocks until there is an active master and that master has completed
272 * initialization.
273 *
274 * @return true if an active master becomes available. false if there are no
275 * masters left.
276 * @throws IOException if something goes wrong or timeout occurs
277 */
278 public boolean waitForActiveAndReadyMaster()
279 throws IOException {
280 return waitForActiveAndReadyMaster(Long.MAX_VALUE);
281 }
282
283 /**
284 * Blocks until there is an active master and that master has completed
285 * initialization.
286 * @param timeout the timeout limit in ms
287 * @return true if an active master becomes available. false if there are no
288 * masters left.
289 */
290 public abstract boolean waitForActiveAndReadyMaster(long timeout)
291 throws IOException;
292
293 /**
294 * Wait for HBase Cluster to shut down.
295 */
296 public abstract void waitUntilShutDown() throws IOException;
297
298 /**
299 * Shut down the HBase cluster
300 */
301 public abstract void shutdown() throws IOException;
302
303 /**
304 * Restores the cluster to it's initial state if this is a real cluster,
305 * otherwise does nothing.
306 * This is a best effort restore. If the servers are not reachable, or insufficient
307 * permissions, etc. restoration might be partial.
308 * @return whether restoration is complete
309 */
310 public boolean restoreInitialStatus() throws IOException {
311 return restoreClusterStatus(getInitialClusterStatus());
312 }
313
314 /**
315 * Restores the cluster to given state if this is a real cluster,
316 * otherwise does nothing.
317 * This is a best effort restore. If the servers are not reachable, or insufficient
318 * permissions, etc. restoration might be partial.
319 * @return whether restoration is complete
320 */
321 public boolean restoreClusterStatus(ClusterStatus desiredStatus) throws IOException {
322 return true;
323 }
324
325 /**
326 * Get the ServerName of region server serving the first hbase:meta region
327 */
328 public ServerName getServerHoldingMeta() throws IOException {
329 return getServerHoldingRegion(TableName.META_TABLE_NAME,
330 HRegionInfo.FIRST_META_REGIONINFO.getRegionName());
331 }
332
333 /**
334 * Get the ServerName of region server serving the specified region
335 * @param regionName Name of the region in bytes
336 * @param tn Table name that has the region.
337 * @return ServerName that hosts the region or null
338 */
339 public abstract ServerName getServerHoldingRegion(final TableName tn, byte[] regionName)
340 throws IOException;
341
342 /**
343 * @return whether we are interacting with a distributed cluster as opposed to an
344 * in-process mini/local cluster.
345 */
346 public boolean isDistributedCluster() {
347 return false;
348 }
349
350 /**
351 * Closes all the resources held open for this cluster. Note that this call does not shutdown
352 * the cluster.
353 * @see #shutdown()
354 */
355 @Override
356 public abstract void close() throws IOException;
357
358 /**
359 * Wait for the namenode.
360 *
361 * @throws InterruptedException
362 */
363 public void waitForNamenodeAvailable() throws InterruptedException {
364 }
365
366 public void waitForDatanodesRegistered(int nbDN) throws Exception {
367 }
368 }