forked from StarRocks/starrocks
-
Notifications
You must be signed in to change notification settings - Fork 1
/
BDBHA.java
290 lines (266 loc) · 12.7 KB
/
BDBHA.java
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
// This file is made available under Elastic License 2.0.
// This file is based on code available under the Apache license here:
// https://github.com/apache/incubator-doris/blob/master/fe/fe-core/src/main/java/org/apache/doris/ha/BDBHA.java
// Licensed to the Apache Software Foundation (ASF) under one
// or more contributor license agreements. See the NOTICE file
// distributed with this work for additional information
// regarding copyright ownership. The ASF licenses this file
// to you under the Apache License, Version 2.0 (the
// "License"); you may not use this file except in compliance
// with the License. You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing,
// software distributed under the License is distributed on an
// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
// KIND, either express or implied. See the License for the
// specific language governing permissions and limitations
// under the License.
package com.starrocks.ha;
import com.google.common.collect.Lists;
import com.google.common.collect.Sets;
import com.sleepycat.bind.tuple.TupleBinding;
import com.sleepycat.je.DatabaseEntry;
import com.sleepycat.je.OperationStatus;
import com.sleepycat.je.rep.MasterStateException;
import com.sleepycat.je.rep.MemberNotFoundException;
import com.sleepycat.je.rep.ReplicatedEnvironment;
import com.sleepycat.je.rep.ReplicationGroup;
import com.sleepycat.je.rep.ReplicationMutableConfig;
import com.sleepycat.je.rep.ReplicationNode;
import com.sleepycat.je.rep.UnknownMasterException;
import com.sleepycat.je.rep.util.ReplicationGroupAdmin;
import com.starrocks.journal.bdbje.BDBEnvironment;
import com.starrocks.journal.bdbje.CloseSafeDatabase;
import com.starrocks.server.GlobalStateMgr;
import org.apache.logging.log4j.LogManager;
import org.apache.logging.log4j.Logger;
import java.net.InetSocketAddress;
import java.util.ArrayList;
import java.util.HashSet;
import java.util.List;
import java.util.Set;
public class BDBHA implements HAProtocol {
private static final Logger LOG = LogManager.getLogger(BDBHA.class);
private final BDBEnvironment environment;
private final String nodeName;
private static final int RETRY_TIME = 3;
// Unstable node is a follower node that is joining the cluster but have not completed.
// We should record this kind node and set the bdb electable group size to
// (size_of_all_followers - size_of_unstable_nodes).
// Because once the handshake is successful, the joined node is put into the optional group,
// but it may take a little time for this node to replicate the historical data.
// This node will never respond to a new data replication until the historical replication is completed,
// and if the master cannot receive a quorum response, the write operation will fail.
private final Set<String> unstableNodes = new HashSet<>();
public BDBHA(BDBEnvironment env, String nodeName) {
this.environment = env;
this.nodeName = nodeName;
}
@Override
public long getEpochNumber() {
return 0;
}
@Override
public boolean fencing() {
CloseSafeDatabase epochDb = environment.getEpochDB();
for (int i = 0; i < RETRY_TIME; i++) {
try {
long count = epochDb.getDb().count();
long myEpoch = count + 1;
LOG.info("start fencing, epoch number is {}", myEpoch);
Long key = myEpoch;
DatabaseEntry theKey = new DatabaseEntry();
TupleBinding<Long> idBinding = TupleBinding.getPrimitiveBinding(Long.class);
idBinding.objectToEntry(key, theKey);
DatabaseEntry theData = new DatabaseEntry(new byte[1]);
OperationStatus status = epochDb.putNoOverwrite(null, theKey, theData);
if (status == OperationStatus.SUCCESS) {
GlobalStateMgr.getCurrentState().setEpoch(myEpoch);
return true;
} else if (status == OperationStatus.KEYEXIST) {
return false;
} else {
throw new Exception(status.toString());
}
} catch (Exception e) {
LOG.error("fencing failed. tried {} times", i, e);
try {
Thread.sleep(2000);
} catch (InterruptedException e1) {
e1.printStackTrace();
}
}
}
return false;
}
@Override
public List<InetSocketAddress> getObserverNodes() {
ReplicationGroupAdmin replicationGroupAdmin = environment.getReplicationGroupAdmin();
if (replicationGroupAdmin == null) {
return null;
}
List<InetSocketAddress> ret = new ArrayList<InetSocketAddress>();
try {
ReplicationGroup replicationGroup = replicationGroupAdmin.getGroup();
for (ReplicationNode replicationNode : replicationGroup.getSecondaryNodes()) {
ret.add(replicationNode.getSocketAddress());
}
} catch (UnknownMasterException e) {
LOG.warn("Catch UnknownMasterException when calling getObserverNodes.", e);
return Lists.newArrayList();
}
return ret;
}
@Override
public List<InetSocketAddress> getElectableNodes(boolean leaderIncluded) {
ReplicationGroupAdmin replicationGroupAdmin = environment.getReplicationGroupAdmin();
if (replicationGroupAdmin == null) {
return null;
}
List<InetSocketAddress> ret = new ArrayList<InetSocketAddress>();
try {
ReplicationGroup replicationGroup = replicationGroupAdmin.getGroup();
for (ReplicationNode replicationNode : replicationGroup.getElectableNodes()) {
if (leaderIncluded) {
ret.add(replicationNode.getSocketAddress());
} else {
if (!replicationNode.getName().equals(replicationGroupAdmin.getMasterNodeName())) {
ret.add(replicationNode.getSocketAddress());
}
}
}
} catch (UnknownMasterException e) {
LOG.warn("Catch UnknownMasterException when calling getElectableNodes.", e);
return Lists.newArrayList();
}
return ret;
}
@Override
public InetSocketAddress getLeader() {
ReplicationGroupAdmin replicationGroupAdmin = environment.getReplicationGroupAdmin();
String leaderName = replicationGroupAdmin.getMasterNodeName();
ReplicationGroup rg = replicationGroupAdmin.getGroup();
ReplicationNode rn = rg.getMember(leaderName);
return rn.getSocketAddress();
}
@Override
public List<InetSocketAddress> getNoneLeaderNodes() {
ReplicationGroupAdmin replicationGroupAdmin = environment.getReplicationGroupAdmin();
if (replicationGroupAdmin == null) {
return null;
}
List<InetSocketAddress> ret = new ArrayList<InetSocketAddress>();
try {
ReplicationGroup replicationGroup = replicationGroupAdmin.getGroup();
for (ReplicationNode replicationNode : replicationGroup.getSecondaryNodes()) {
ret.add(replicationNode.getSocketAddress());
}
for (ReplicationNode replicationNode : replicationGroup.getElectableNodes()) {
if (!replicationNode.getName().equals(replicationGroupAdmin.getMasterNodeName())) {
ret.add(replicationNode.getSocketAddress());
}
}
} catch (UnknownMasterException e) {
LOG.warn("Catch UnknownMasterException when calling getNoneLeaderNodes.", e);
return null;
}
return ret;
}
@Override
public boolean isLeader() {
ReplicationGroupAdmin replicationGroupAdmin = environment.getReplicationGroupAdmin();
String leaderName = replicationGroupAdmin.getMasterNodeName();
return leaderName.equals(nodeName);
}
@Override
public boolean removeElectableNode(String nodeName) {
ReplicationGroupAdmin replicationGroupAdmin = environment.getReplicationGroupAdmin();
if (replicationGroupAdmin == null) {
return false;
}
try {
replicationGroupAdmin.removeMember(nodeName);
} catch (MemberNotFoundException e) {
LOG.error("the deleting electable node is not found {}", nodeName, e);
return false;
} catch (MasterStateException e) {
LOG.error("the deleting electable node is master {}", nodeName, e);
return false;
}
return true;
}
// When new Follower FE is added to the cluster, it should also be added to the helper sockets in
// ReplicationGroupAdmin, in order to fix the following case:
// 1. A Observer starts with helper of master FE.
// 2. Master FE is dead, new Master is elected.
// 3. Observer's helper sockets only contains the info of the dead master FE.
// So when you try to get frontends' info from this Observer, it will throw the Exception:
// "Could not determine master from helpers at:[/dead master FE host:port]"
public void addHelperSocket(String ip, Integer port) {
ReplicationGroupAdmin replicationGroupAdmin = environment.getReplicationGroupAdmin();
Set<InetSocketAddress> helperSockets = Sets.newHashSet(replicationGroupAdmin.getHelperSockets());
InetSocketAddress newHelperSocket = new InetSocketAddress(ip, port);
if (!helperSockets.contains(newHelperSocket)) {
helperSockets.add(newHelperSocket);
environment.setNewReplicationGroupAdmin(helperSockets);
LOG.info("add {}:{} to helper sockets", ip, port);
}
}
public void removeHelperSocket(String ip, Integer port) {
ReplicationGroupAdmin replicationGroupAdmin = environment.getReplicationGroupAdmin();
Set<InetSocketAddress> helperSockets = Sets.newHashSet(replicationGroupAdmin.getHelperSockets());
InetSocketAddress targetAddress = new InetSocketAddress(ip, port);
if (helperSockets.contains(targetAddress)) {
helperSockets.remove(targetAddress);
environment.setNewReplicationGroupAdmin(helperSockets);
LOG.info("remove helper socket {}:{} from replicationGroupAdmin", ip, port);
}
}
public void removeNodeIfExist(String host, int port) {
ReplicationGroupAdmin replicationGroupAdmin = environment.getReplicationGroupAdmin();
if (replicationGroupAdmin == null) {
return;
}
List<String> conflictNodes = Lists.newArrayList();
Set<ReplicationNode> nodes = replicationGroupAdmin.getGroup().getElectableNodes();
for (ReplicationNode node : nodes) {
if (node.getHostName().equals(host) && node.getPort() == port) {
conflictNodes.add(node.getName());
}
}
for (String nodeName : conflictNodes) {
removeElectableNode(nodeName);
}
}
public synchronized void addUnstableNode(String nodeName, int currentFollowerCnt) {
unstableNodes.add(nodeName);
ReplicatedEnvironment replicatedEnvironment = environment.getReplicatedEnvironment();
if (replicatedEnvironment != null) {
replicatedEnvironment.
setRepMutableConfig(new ReplicationMutableConfig().
setElectableGroupSizeOverride(currentFollowerCnt - unstableNodes.size()));
}
}
public synchronized void removeUnstableNode(String nodeName, int currentFollowerCnt) {
unstableNodes.remove(nodeName);
ReplicatedEnvironment replicatedEnvironment = environment.getReplicatedEnvironment();
if (replicatedEnvironment != null) {
if (unstableNodes.isEmpty()) {
// Setting ElectableGroupSizeOverride to 0 means remove this config,
// and bdb will use the normal electable group size.
replicatedEnvironment.
setRepMutableConfig(new ReplicationMutableConfig().setElectableGroupSizeOverride(0));
} else {
replicatedEnvironment.
setRepMutableConfig(new ReplicationMutableConfig().
setElectableGroupSizeOverride(currentFollowerCnt - unstableNodes.size()));
}
}
}
public synchronized void updateFrontendHostAndPort(String nodeName, String newHostName, int newPort) {
ReplicationGroupAdmin replicationGroupAdmin = environment.getReplicationGroupAdmin();
replicationGroupAdmin.updateAddress(nodeName, newHostName, newPort);
}
}