1 /**
2 *
3 * Licensed to the Apache Software Foundation (ASF) under one
4 * or more contributor license agreements. See the NOTICE file
5 * distributed with this work for additional information
6 * regarding copyright ownership. The ASF licenses this file
7 * to you under the Apache License, Version 2.0 (the
8 * "License"); you may not use this file except in compliance
9 * with the License. You may obtain a copy of the License at
10 *
11 * http://www.apache.org/licenses/LICENSE-2.0
12 *
13 * Unless required by applicable law or agreed to in writing, software
14 * distributed under the License is distributed on an "AS IS" BASIS,
15 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
16 * See the License for the specific language governing permissions and
17 * limitations under the License.
18 */
19
20 package org.apache.hadoop.hbase.coordination;
21
22 import java.io.IOException;
23 import java.io.InterruptedIOException;
24 import java.util.Set;
25 import java.util.concurrent.ConcurrentMap;
26
27 import org.apache.hadoop.hbase.classification.InterfaceAudience;
28 import org.apache.hadoop.hbase.HRegionInfo;
29 import org.apache.hadoop.hbase.ServerName;
30 import org.apache.hadoop.hbase.master.MasterServices;
31 import org.apache.hadoop.hbase.master.SplitLogManager.ResubmitDirective;
32 import org.apache.hadoop.hbase.master.SplitLogManager.Task;
33 import org.apache.hadoop.hbase.protobuf.generated.ZooKeeperProtos.SplitLogTask.RecoveryMode;
34
35 import com.google.common.annotations.VisibleForTesting;
36
37 /**
38 * Coordination for SplitLogManager. It creates and works with tasks for split log operations<BR>
39 * Manager prepares task by calling {@link #prepareTask} and submit it by
40 * {@link #submitTask(String)}. After that it periodically check the number of remaining tasks by
41 * {@link #remainingTasksInCoordination()} and waits until it become zero.
42 * <P>
43 * Methods required for task life circle: <BR>
44 * {@link #markRegionsRecovering(ServerName, Set)} mark regions for log replaying. Used by
45 * {@link org.apache.hadoop.hbase.master.MasterFileSystem} <BR>
46 * {@link #removeRecoveringRegions(Set, Boolean)} make regions cleanup that previous were marked as
47 * recovering. Called after all tasks processed <BR>
48 * {@link #removeStaleRecoveringRegions(Set)} remove stale recovering. called by
49 * {@link org.apache.hadoop.hbase.master.MasterFileSystem} after Active Master is initialized <BR>
50 * {@link #getLastRecoveryTime()} required for garbage collector and should indicate when the last
51 * recovery has been made<BR>
52 * {@link #checkTaskStillAvailable(String)} Check that task is still there <BR>
53 * {@link #checkTasks()} check for unassigned tasks and resubmit them
54 */
55 @InterfaceAudience.Private
56 public interface SplitLogManagerCoordination {
57
58 /**
59 * Detail class that shares data between coordination and split log manager
60 */
61 public static class SplitLogManagerDetails {
62 final private ConcurrentMap<String, Task> tasks;
63 final private MasterServices master;
64 final private Set<String> failedDeletions;
65 final private ServerName serverName;
66
67 public SplitLogManagerDetails(ConcurrentMap<String, Task> tasks, MasterServices master,
68 Set<String> failedDeletions, ServerName serverName) {
69 this.tasks = tasks;
70 this.master = master;
71 this.failedDeletions = failedDeletions;
72 this.serverName = serverName;
73 }
74
75 /**
76 * @return the master value
77 */
78 public MasterServices getMaster() {
79 return master;
80 }
81
82 /**
83 * @return map of tasks
84 */
85 public ConcurrentMap<String, Task> getTasks() {
86 return tasks;
87 }
88
89 /**
90 * @return a set of failed deletions
91 */
92 public Set<String> getFailedDeletions() {
93 return failedDeletions;
94 }
95
96 /**
97 * @return server name
98 */
99 public ServerName getServerName() {
100 return serverName;
101 }
102 }
103
104 /**
105 * Provide the configuration from the SplitLogManager
106 */
107 void setDetails(SplitLogManagerDetails details);
108
109 /**
110 * Returns the configuration that was provided previously
111 */
112 SplitLogManagerDetails getDetails();
113
114 /**
115 * Prepare the new task
116 * @param taskName name of the task
117 * @return the task id
118 */
119 String prepareTask(String taskName);
120
121 /**
122 * Mark regions in recovering state for distributed log replay
123 * @param serverName server name
124 * @param userRegions set of regions to be marked
125 * @throws IOException in case of failure
126 * @throws InterruptedIOException
127 */
128 void markRegionsRecovering(final ServerName serverName, Set<HRegionInfo> userRegions)
129 throws IOException, InterruptedIOException;
130
131 /**
132 * tells Coordination that it should check for new tasks
133 */
134 void checkTasks();
135
136 /**
137 * It removes recovering regions from Coordination
138 * @param serverNames servers which are just recovered
139 * @param isMetaRecovery whether current recovery is for the meta region on
140 * <code>serverNames</code>
141 */
142 void removeRecoveringRegions(Set<String> serverNames, Boolean isMetaRecovery) throws IOException;
143
144 /**
145 * Return the number of remaining tasks
146 */
147 int remainingTasksInCoordination();
148
149 /**
150 * Check that the task is still there
151 * @param task node to check
152 */
153 void checkTaskStillAvailable(String task);
154
155 /**
156 * Change the recovery mode.
157 * @param b the recovery mode state
158 * @throws InterruptedIOException
159 * @throws IOException in case of failure
160 */
161 void setRecoveryMode(boolean b) throws InterruptedIOException, IOException;
162
163 /**
164 * Removes known stale servers
165 * @param knownServers set of previously failed servers
166 * @throws IOException in case of failure
167 * @throws InterruptedIOException
168 */
169 void removeStaleRecoveringRegions(Set<String> knownServers) throws IOException,
170 InterruptedIOException;
171
172 /**
173 * Resubmit the task in case if found unassigned or failed
174 * @param taskName path related to task
175 * @param task to resubmit
176 * @param force whether it should be forced
177 * @return whether it was successful
178 */
179
180 boolean resubmitTask(String taskName, Task task, ResubmitDirective force);
181
182 /**
183 * @param taskName to be submitted
184 */
185 void submitTask(String taskName);
186
187 /**
188 * @param taskName to be removed
189 */
190 void deleteTask(String taskName);
191
192 /**
193 * @return shows whether the log recovery mode is in replaying state
194 */
195 boolean isReplaying();
196
197 /**
198 * @return shows whether the log recovery mode is in splitting state
199 */
200 boolean isSplitting();
201
202 /**
203 * @return the time of last attempt to recover
204 */
205 long getLastRecoveryTime();
206
207 /**
208 * Temporary function, mostly for UTs. In the regular code isReplaying or isSplitting should be
209 * used.
210 * @return the current log recovery mode.
211 */
212 RecoveryMode getRecoveryMode();
213
214 /**
215 * Support method to init constants such as timeout. Mostly required for UTs.
216 * @throws IOException
217 */
218 @VisibleForTesting
219 void init() throws IOException;
220 }