-
Notifications
You must be signed in to change notification settings - Fork 1.8k
/
AbstractBlockingService.java
426 lines (363 loc) · 15.9 KB
/
AbstractBlockingService.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
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
/*
* Copyright (c) 2008-2024, Hazelcast, Inc. All Rights Reserved.
*
* Licensed 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.hazelcast.cp.internal.datastructures.spi.blocking;
import com.hazelcast.cluster.Address;
import com.hazelcast.cp.CPGroupId;
import com.hazelcast.cp.internal.RaftNodeLifecycleAwareService;
import com.hazelcast.cp.internal.RaftService;
import com.hazelcast.cp.internal.datastructures.spi.AbstractCPMigrationAwareService;
import com.hazelcast.cp.internal.datastructures.spi.RaftManagedService;
import com.hazelcast.cp.internal.datastructures.spi.RaftRemoteService;
import com.hazelcast.cp.internal.datastructures.spi.blocking.operation.ExpireWaitKeysOp;
import com.hazelcast.cp.internal.operation.unsafe.UnsafeRaftReplicateOp;
import com.hazelcast.cp.internal.raft.SnapshotAwareService;
import com.hazelcast.cp.internal.raft.impl.RaftNode;
import com.hazelcast.cp.internal.session.SessionAccessor;
import com.hazelcast.cp.internal.session.SessionAwareService;
import com.hazelcast.cp.internal.session.SessionExpiredException;
import com.hazelcast.internal.partition.MigrationAwareService;
import com.hazelcast.internal.util.BiTuple;
import com.hazelcast.internal.util.Clock;
import com.hazelcast.internal.util.collection.Long2ObjectHashMap;
import com.hazelcast.logging.ILogger;
import com.hazelcast.spi.exception.DistributedObjectDestroyedException;
import com.hazelcast.spi.impl.InternalCompletableFuture;
import com.hazelcast.spi.impl.NodeEngine;
import com.hazelcast.spi.impl.executionservice.ExecutionService;
import com.hazelcast.spi.impl.operationservice.LiveOperations;
import com.hazelcast.spi.impl.operationservice.LiveOperationsTracker;
import java.util.ArrayList;
import java.util.Collection;
import java.util.Collections;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import java.util.Map.Entry;
import java.util.Properties;
import java.util.Set;
import java.util.UUID;
import java.util.concurrent.ConcurrentHashMap;
import java.util.concurrent.ConcurrentMap;
import java.util.stream.Collectors;
import static com.hazelcast.cp.internal.session.AbstractProxySessionManager.NO_SESSION_ID;
import static com.hazelcast.internal.util.Preconditions.checkNotNull;
import static java.lang.Thread.currentThread;
import static java.util.concurrent.TimeUnit.MILLISECONDS;
/**
* Base class for services that maintain blocking resources.
* Contains common behaviour that will be needed by service implementations.
*
* @param <W> concrete type of the WaitKey
* @param <R> concrete type of the resource
* @param <RR> concrete ty;e lf the resource registry
*/
@SuppressWarnings("checkstyle:methodcount")
public abstract class AbstractBlockingService<W extends WaitKey, R extends BlockingResource<W>, RR extends ResourceRegistry<W, R>>
extends AbstractCPMigrationAwareService
implements RaftManagedService, RaftNodeLifecycleAwareService, RaftRemoteService, SessionAwareService,
SnapshotAwareService<RR>, LiveOperationsTracker, MigrationAwareService {
public static final long WAIT_TIMEOUT_TASK_UPPER_BOUND_MILLIS = 1500;
private static final long WAIT_TIMEOUT_TASK_PERIOD_MILLIS = 500;
protected final ILogger logger;
protected volatile RaftService raftService;
private final ConcurrentMap<CPGroupId, RR> registries = new ConcurrentHashMap<>();
private volatile SessionAccessor sessionAccessor;
protected AbstractBlockingService(NodeEngine nodeEngine) {
super(nodeEngine);
this.logger = nodeEngine.getLogger(getClass());
}
@Override
public final void init(NodeEngine nodeEngine, Properties properties) {
this.raftService = nodeEngine.getService(RaftService.SERVICE_NAME);
ExecutionService executionService = nodeEngine.getExecutionService();
executionService.scheduleWithRepetition(new ExpireWaitKeysPeriodicTask(),
WAIT_TIMEOUT_TASK_PERIOD_MILLIS, WAIT_TIMEOUT_TASK_PERIOD_MILLIS, MILLISECONDS);
initImpl();
}
/**
* Subclasses can implement their custom initialization logic here
*/
protected void initImpl() {
}
@Override
public void reset() {
if (!raftService.isCpSubsystemEnabled()) {
registries.clear();
}
}
@Override
public void onCPSubsystemRestart() {
registries.clear();
}
@Override
public final void shutdown(boolean terminate) {
registries.clear();
shutdownImpl(terminate);
}
/**
* Subclasses can implement their custom shutdown logic here
*/
protected void shutdownImpl(boolean terminate) {
}
/**
* Returns name of the service.
*/
protected abstract String serviceName();
/**
* Creates a registry for the given Raft group.
*/
protected abstract RR createNewRegistry(CPGroupId groupId);
/**
* Creates the response object that will be sent for a expired wait key.
*/
protected abstract Object expiredWaitKeyResponse();
protected void onRegistryRestored(RR registry) {
}
@Override
public boolean destroyRaftObject(CPGroupId groupId, String name) {
Collection<W> keys = getOrInitRegistry(groupId).destroyResource(name);
if (keys == null) {
return false;
}
List<Long> commitIndices = new ArrayList<>();
for (W key : keys) {
commitIndices.add(key.commitIndex());
}
completeFutures(groupId, commitIndices, new DistributedObjectDestroyedException(name + " is destroyed"));
return true;
}
@Override
public final RR takeSnapshot(CPGroupId groupId, long commitIndex) {
RR registry = getRegistryOrNull(groupId);
return registry != null ? (RR) registry.cloneForSnapshot() : null;
}
@Override
public final void restoreSnapshot(CPGroupId groupId, long commitIndex, RR registry) {
RR prev = registries.put(registry.getGroupId(), registry);
// do not shift the already existing wait timeouts...
Map<BiTuple<String, UUID>, BiTuple<Long, Long>> existingWaitTimeouts =
prev != null ? prev.getWaitTimeouts() : Collections.emptyMap();
Map<BiTuple<String, UUID>, Long> newWaitKeys = registry.overwriteWaitTimeouts(existingWaitTimeouts);
for (Entry<BiTuple<String, UUID>, Long> e : newWaitKeys.entrySet()) {
scheduleTimeout(groupId, e.getKey().element1, e.getKey().element2, e.getValue());
}
registry.onSnapshotRestore();
onRegistryRestored(registry);
}
@Override
public void setSessionAccessor(SessionAccessor accessor) {
this.sessionAccessor = accessor;
}
@Override
public final void onSessionClose(CPGroupId groupId, long sessionId) {
ResourceRegistry<W, R> registry = registries.get(groupId);
if (registry == null) {
if (logger.isFineEnabled()) {
logger.fine("Resource registry of " + groupId + " not found to handle closed Session[" + sessionId + "]");
}
return;
}
List<Long> expiredWaitKeys = new ArrayList<>();
Long2ObjectHashMap<Object> completedWaitKeys = new Long2ObjectHashMap<>();
registry.closeSession(sessionId, expiredWaitKeys, completedWaitKeys);
if (logger.isFineEnabled() && !(expiredWaitKeys.isEmpty() && completedWaitKeys.isEmpty())) {
logger.fine("Closed Session[" + sessionId + "] in " + groupId + " expired wait key commit indices: "
+ expiredWaitKeys + " completed wait keys: " + completedWaitKeys);
}
completeFutures(groupId, expiredWaitKeys, new SessionExpiredException());
raftService.completeFutures(groupId, completedWaitKeys.entrySet());
}
@Override
public final Collection<Long> getAttachedSessions(CPGroupId groupId) {
RR registry = getRegistryOrNull(groupId);
return registry != null ? registry.getAttachedSessions() : Collections.emptyList();
}
@Override
public final void onRaftNodeTerminated(CPGroupId groupId) {
ResourceRegistry<W, R> registry = registries.get(groupId);
if (registry != null) {
Collection<Long> indices = registry.destroy();
completeFutures(groupId, indices, new DistributedObjectDestroyedException(groupId + " is destroyed"));
}
}
@Override
public final void onRaftNodeSteppedDown(CPGroupId groupId) {
}
@Override
public final void populate(LiveOperations liveOperations) {
long now = Clock.currentTimeMillis();
for (RR registry : registries.values()) {
registry.populate(liveOperations, now);
}
}
public final void expireWaitKeys(CPGroupId groupId, Collection<BiTuple<String, UUID>> keys) {
// no need to validate the session. if the session is expired, the corresponding wait key is gone already
ResourceRegistry<W, R> registry = registries.get(groupId);
if (registry == null) {
logger.severe("Registry of " + groupId + " not found to expire wait keys: " + keys);
return;
}
List<W> expired = new ArrayList<>();
for (BiTuple<String, UUID> key : keys) {
registry.expireWaitKey(key.element1, key.element2, expired);
}
List<Long> commitIndices = new ArrayList<>();
for (W key : expired) {
commitIndices.add(key.commitIndex());
registry.removeLiveOperation(key);
}
completeFutures(groupId, commitIndices, expiredWaitKeyResponse());
}
public final RR getRegistryOrNull(CPGroupId groupId) {
return registries.get(groupId);
}
public Collection<BiTuple<Address, Long>> getLiveOperations(CPGroupId groupId) {
RR registry = registries.get(groupId);
if (registry == null) {
return Collections.emptySet();
}
return Collections.unmodifiableCollection(registry.getLiveOperations());
}
public int getTotalResourcesCount() {
return registries.values().stream()
.mapToInt(collection -> collection.resources.size())
.sum();
}
// squid:S3824 ConcurrentHashMap.computeIfAbsent(K, Function<? super K, ? extends V>) locks the map, which *may* have an
// effect on throughput such that it's not a direct replacement
@SuppressWarnings("squid:S3824")
protected final RR getOrInitRegistry(CPGroupId groupId) {
checkNotNull(groupId);
RR registry = registries.get(groupId);
if (registry == null) {
registry = createNewRegistry(groupId);
registries.put(groupId, registry);
}
return registry;
}
protected final void scheduleTimeout(CPGroupId groupId, String name, UUID invocationUid, long timeoutMs) {
if (timeoutMs > 0 && timeoutMs <= WAIT_TIMEOUT_TASK_UPPER_BOUND_MILLIS) {
ExecutionService executionService = nodeEngine.getExecutionService();
executionService.schedule(new ExpireWaitKeysTask(groupId, BiTuple.of(name, invocationUid)), timeoutMs, MILLISECONDS);
}
}
protected final void heartbeatSession(CPGroupId groupId, long sessionId) {
if (sessionId == NO_SESSION_ID) {
return;
}
if (sessionAccessor.isActive(groupId, sessionId)) {
sessionAccessor.heartbeat(groupId, sessionId);
return;
}
throw new SessionExpiredException("active session: " + sessionId + " does not exist in " + groupId);
}
protected final void notifyWaitKeys(CPGroupId groupId, String name, Collection<W> keys, Object result) {
if (keys.isEmpty()) {
return;
}
if (logger.isFineEnabled()) {
logger.fine("Resource[" + name + "] in " + groupId + " completed wait keys: " + keys + " result: " + result);
}
List<Long> indices = new ArrayList<>(keys.size());
for (W key : keys) {
indices.add(key.commitIndex());
}
completeFutures(groupId, indices, result);
}
private void completeFutures(CPGroupId groupId, Collection<Long> indices, Object result) {
if (!indices.isEmpty()) {
if (!raftService.completeFutures(groupId, indices, result)) {
logger.severe("RaftNode not found for " + groupId + " to notify commit indices " + indices + " with " + result);
}
}
}
private void tryReplicateExpiredWaitKeys(CPGroupId groupId, Collection<BiTuple<String, UUID>> keys) {
InternalCompletableFuture future = null;
try {
ExpireWaitKeysOp op = new ExpireWaitKeysOp(serviceName(), keys);
if (raftService.isCpSubsystemEnabled()) {
RaftNode raftNode = raftService.getRaftNode(groupId);
if (raftNode != null) {
future = raftNode.replicate(op);
}
} else {
future = raftService.getInvocationManager()
.invokeOnPartition(new UnsafeRaftReplicateOp(groupId, op));
}
if (future != null) {
future.get(WAIT_TIMEOUT_TASK_PERIOD_MILLIS, MILLISECONDS);
}
} catch (Exception e) {
if (logger.isFineEnabled()) {
logger.fine("Could not expire wait keys: " + keys + " in " + groupId, e);
}
}
}
protected Set<CPGroupId> getGroupIdSet() {
return registries.keySet();
}
private class ExpireWaitKeysTask implements Runnable {
final CPGroupId groupId;
final Collection<BiTuple<String, UUID>> keys;
ExpireWaitKeysTask(CPGroupId groupId, BiTuple<String, UUID> key) {
this.groupId = groupId;
this.keys = Collections.singleton(key);
}
@Override
public void run() {
tryReplicateExpiredWaitKeys(groupId, keys);
}
}
private class ExpireWaitKeysPeriodicTask implements Runnable {
@Override
public void run() {
for (Entry<CPGroupId, Collection<BiTuple<String, UUID>>> e : getWaitKeysToExpire().entrySet()) {
if (currentThread().isInterrupted()) {
break;
}
tryReplicateExpiredWaitKeys(e.getKey(), e.getValue());
}
}
// queried locally
private Map<CPGroupId, Collection<BiTuple<String, UUID>>> getWaitKeysToExpire() {
Map<CPGroupId, Collection<BiTuple<String, UUID>>> timeouts = new HashMap<>();
long now = Clock.currentTimeMillis();
for (ResourceRegistry<W, R> registry : registries.values()) {
Collection<BiTuple<String, UUID>> t = registry.getWaitKeysToExpire(now);
if (!t.isEmpty()) {
timeouts.put(registry.getGroupId(), t);
}
}
return timeouts;
}
}
@Override
protected int getBackupCount() {
return 1;
}
@Override
protected Map<CPGroupId, Object> getSnapshotMap(int partitionId) {
assert !raftService.isCpSubsystemEnabled();
return getGroupIdSet().stream()
.filter(groupId -> raftService.getCPGroupPartitionId(groupId) == partitionId)
.map(groupId -> BiTuple.of(groupId, takeSnapshot(groupId, 0L)))
.collect(Collectors.toMap(tuple -> tuple.element1, tuple -> tuple.element2));
}
@Override
protected void clearPartitionReplica(int partitionId) {
getGroupIdSet().removeIf(groupId -> raftService.getCPGroupPartitionId(groupId) == partitionId);
}
}