aboutsummaryrefslogtreecommitdiffstats
path: root/framework/src/onos/core/store/dist/src/main/java/org/onosproject/store/flow/impl/NewDistributedFlowRuleStore.java
blob: 1695e5ff576cd503305fd4d1dfdaad0d242e14e6 (plain)
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
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
 /*
 * Copyright 2014-2015 Open Networking Laboratory
 *
 * 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 org.onosproject.store.flow.impl;

import com.google.common.base.Objects;
import com.google.common.collect.ImmutableList;
import com.google.common.collect.ImmutableMap;
import com.google.common.collect.Iterables;
import com.google.common.collect.Maps;
import com.google.common.collect.Sets;
import com.google.common.util.concurrent.Futures;

import org.apache.felix.scr.annotations.Activate;
import org.apache.felix.scr.annotations.Component;
import org.apache.felix.scr.annotations.Deactivate;
import org.apache.felix.scr.annotations.Modified;
import org.apache.felix.scr.annotations.Property;
import org.apache.felix.scr.annotations.Reference;
import org.apache.felix.scr.annotations.ReferenceCardinality;
import org.apache.felix.scr.annotations.Service;
import org.onlab.util.KryoNamespace;
import org.onlab.util.Tools;
import org.onosproject.cfg.ComponentConfigService;
import org.onosproject.cluster.ClusterService;
import org.onosproject.cluster.NodeId;
import org.onosproject.core.CoreService;
import org.onosproject.core.IdGenerator;
import org.onosproject.mastership.MastershipService;
import org.onosproject.net.DeviceId;
import org.onosproject.net.device.DeviceService;
import org.onosproject.net.flow.CompletedBatchOperation;
import org.onosproject.net.flow.DefaultFlowEntry;
import org.onosproject.net.flow.FlowEntry;
import org.onosproject.net.flow.FlowEntry.FlowEntryState;
import org.onosproject.net.flow.FlowId;
import org.onosproject.net.flow.FlowRule;
import org.onosproject.net.flow.FlowRuleBatchEntry;
import org.onosproject.net.flow.FlowRuleBatchEntry.FlowRuleOperation;
import org.onosproject.net.flow.FlowRuleBatchEvent;
import org.onosproject.net.flow.FlowRuleBatchOperation;
import org.onosproject.net.flow.FlowRuleBatchRequest;
import org.onosproject.net.flow.FlowRuleEvent;
import org.onosproject.net.flow.FlowRuleEvent.Type;
import org.onosproject.net.flow.FlowRuleService;
import org.onosproject.net.flow.FlowRuleStore;
import org.onosproject.net.flow.FlowRuleStoreDelegate;
import org.onosproject.net.flow.StoredFlowEntry;
import org.onosproject.net.flow.TableStatisticsEntry;
import org.onosproject.persistence.PersistenceService;
import org.onosproject.store.AbstractStore;
import org.onosproject.store.cluster.messaging.ClusterCommunicationService;
import org.onosproject.store.cluster.messaging.ClusterMessage;
import org.onosproject.store.cluster.messaging.ClusterMessageHandler;
import org.onosproject.store.flow.ReplicaInfoEvent;
import org.onosproject.store.flow.ReplicaInfoEventListener;
import org.onosproject.store.flow.ReplicaInfoService;
import org.onosproject.store.impl.MastershipBasedTimestamp;
import org.onosproject.store.serializers.KryoNamespaces;
import org.onosproject.store.serializers.KryoSerializer;
import org.onosproject.store.serializers.StoreSerializer;
import org.onosproject.store.serializers.custom.DistributedStoreSerializers;
import org.onosproject.store.service.EventuallyConsistentMap;
import org.onosproject.store.service.EventuallyConsistentMapEvent;
import org.onosproject.store.service.EventuallyConsistentMapListener;
import org.onosproject.store.service.Serializer;
import org.onosproject.store.service.StorageService;
import org.onosproject.store.service.WallClockTimestamp;
import org.osgi.service.component.ComponentContext;
import org.slf4j.Logger;

import java.util.Collections;
import java.util.Dictionary;
import java.util.HashSet;
import java.util.List;
import java.util.Map;
import java.util.Set;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import java.util.concurrent.ScheduledExecutorService;
import java.util.concurrent.ScheduledFuture;
import java.util.concurrent.TimeUnit;
import java.util.concurrent.atomic.AtomicInteger;
import java.util.stream.Collectors;

import static com.google.common.base.Strings.isNullOrEmpty;
import static org.onlab.util.Tools.get;
import static org.onlab.util.Tools.groupedThreads;
import static org.onosproject.net.flow.FlowRuleEvent.Type.RULE_REMOVED;
import static org.onosproject.store.flow.impl.FlowStoreMessageSubjects.*;
import static org.slf4j.LoggerFactory.getLogger;

/**
 * Manages inventory of flow rules using a distributed state management protocol.
 */
@Component(immediate = true, enabled = true)
@Service
public class NewDistributedFlowRuleStore
        extends AbstractStore<FlowRuleBatchEvent, FlowRuleStoreDelegate>
        implements FlowRuleStore {

    private final Logger log = getLogger(getClass());

    private static final int MESSAGE_HANDLER_THREAD_POOL_SIZE = 8;
    private static final boolean DEFAULT_BACKUP_ENABLED = true;
    private static final boolean DEFAULT_PERSISTENCE_ENABLED = false;
    private static final int DEFAULT_BACKUP_PERIOD_MILLIS = 2000;
    private static final long FLOW_RULE_STORE_TIMEOUT_MILLIS = 5000;
    // number of devices whose flow entries will be backed up in one communication round
    private static final int FLOW_TABLE_BACKUP_BATCH_SIZE = 1;

    @Property(name = "msgHandlerPoolSize", intValue = MESSAGE_HANDLER_THREAD_POOL_SIZE,
            label = "Number of threads in the message handler pool")
    private int msgHandlerPoolSize = MESSAGE_HANDLER_THREAD_POOL_SIZE;

    @Property(name = "backupEnabled", boolValue = DEFAULT_BACKUP_ENABLED,
            label = "Indicates whether backups are enabled or not")
    private boolean backupEnabled = DEFAULT_BACKUP_ENABLED;

    @Property(name = "backupPeriod", intValue = DEFAULT_BACKUP_PERIOD_MILLIS,
            label = "Delay in ms between successive backup runs")
    private int backupPeriod = DEFAULT_BACKUP_PERIOD_MILLIS;
    @Property(name = "persistenceEnabled", boolValue = false,
            label = "Indicates whether or not changes in the flow table should be persisted to disk.")
    private boolean persistenceEnabled = DEFAULT_PERSISTENCE_ENABLED;

    private InternalFlowTable flowTable = new InternalFlowTable();

    @Reference(cardinality = ReferenceCardinality.MANDATORY_UNARY)
    protected ReplicaInfoService replicaInfoManager;

    @Reference(cardinality = ReferenceCardinality.MANDATORY_UNARY)
    protected ClusterCommunicationService clusterCommunicator;

    @Reference(cardinality = ReferenceCardinality.MANDATORY_UNARY)
    protected ClusterService clusterService;

    @Reference(cardinality = ReferenceCardinality.MANDATORY_UNARY)
    protected DeviceService deviceService;

    @Reference(cardinality = ReferenceCardinality.MANDATORY_UNARY)
    protected CoreService coreService;

    @Reference(cardinality = ReferenceCardinality.MANDATORY_UNARY)
    protected ComponentConfigService configService;

    @Reference(cardinality = ReferenceCardinality.MANDATORY_UNARY)
    protected MastershipService mastershipService;

    @Reference(cardinality = ReferenceCardinality.MANDATORY_UNARY)
    protected PersistenceService persistenceService;

    private Map<Long, NodeId> pendingResponses = Maps.newConcurrentMap();
    private ExecutorService messageHandlingExecutor;

    private ScheduledFuture<?> backupTask;
    private final ScheduledExecutorService backupSenderExecutor =
            Executors.newSingleThreadScheduledExecutor(groupedThreads("onos/flow", "backup-sender"));

    private EventuallyConsistentMap<DeviceId, List<TableStatisticsEntry>> deviceTableStats;
    private final EventuallyConsistentMapListener<DeviceId, List<TableStatisticsEntry>> tableStatsListener =
            new InternalTableStatsListener();

    @Reference(cardinality = ReferenceCardinality.MANDATORY_UNARY)
    protected StorageService storageService;

    protected static final StoreSerializer SERIALIZER = new KryoSerializer() {
        @Override
        protected void setupKryoPool() {
            serializerPool = KryoNamespace.newBuilder()
                    .register(DistributedStoreSerializers.STORE_COMMON)
                    .nextId(DistributedStoreSerializers.STORE_CUSTOM_BEGIN)
                    .build();
        }
    };

    protected static final KryoNamespace.Builder SERIALIZER_BUILDER = KryoNamespace.newBuilder()
            .register(KryoNamespaces.API)
            .register(MastershipBasedTimestamp.class);


    private IdGenerator idGenerator;
    private NodeId local;

    @Activate
    public void activate(ComponentContext context) {
        configService.registerProperties(getClass());

        idGenerator = coreService.getIdGenerator(FlowRuleService.FLOW_OP_TOPIC);

        local = clusterService.getLocalNode().id();

        messageHandlingExecutor = Executors.newFixedThreadPool(
                msgHandlerPoolSize, groupedThreads("onos/store/flow", "message-handlers"));

        registerMessageHandlers(messageHandlingExecutor);

        if (backupEnabled) {
            replicaInfoManager.addListener(flowTable);
            backupTask = backupSenderExecutor.scheduleWithFixedDelay(
                    flowTable::backup,
                    0,
                    backupPeriod,
                    TimeUnit.MILLISECONDS);
        }

        deviceTableStats = storageService.<DeviceId, List<TableStatisticsEntry>>eventuallyConsistentMapBuilder()
                .withName("onos-flow-table-stats")
                .withSerializer(SERIALIZER_BUILDER)
                .withAntiEntropyPeriod(5, TimeUnit.SECONDS)
                .withTimestampProvider((k, v) -> new WallClockTimestamp())
                .withTombstonesDisabled()
                .build();
        deviceTableStats.addListener(tableStatsListener);

        logConfig("Started");
    }

    @Deactivate
    public void deactivate(ComponentContext context) {
        if (backupEnabled) {
            replicaInfoManager.removeListener(flowTable);
            backupTask.cancel(true);
        }
        configService.unregisterProperties(getClass(), false);
        unregisterMessageHandlers();
        deviceTableStats.removeListener(tableStatsListener);
        deviceTableStats.destroy();
        messageHandlingExecutor.shutdownNow();
        backupSenderExecutor.shutdownNow();
        log.info("Stopped");
    }

    @SuppressWarnings("rawtypes")
    @Modified
    public void modified(ComponentContext context) {
        if (context == null) {
            backupEnabled = DEFAULT_BACKUP_ENABLED;
            logConfig("Default config");
            return;
        }

        Dictionary properties = context.getProperties();
        int newPoolSize;
        boolean newBackupEnabled;
        int newBackupPeriod;
        try {
            String s = get(properties, "msgHandlerPoolSize");
            newPoolSize = isNullOrEmpty(s) ? msgHandlerPoolSize : Integer.parseInt(s.trim());

            s = get(properties, "backupEnabled");
            newBackupEnabled = isNullOrEmpty(s) ? backupEnabled : Boolean.parseBoolean(s.trim());

            s = get(properties, "backupPeriod");
            newBackupPeriod = isNullOrEmpty(s) ? backupPeriod : Integer.parseInt(s.trim());

        } catch (NumberFormatException | ClassCastException e) {
            newPoolSize = MESSAGE_HANDLER_THREAD_POOL_SIZE;
            newBackupEnabled = DEFAULT_BACKUP_ENABLED;
            newBackupPeriod = DEFAULT_BACKUP_PERIOD_MILLIS;
        }

        boolean restartBackupTask = false;
        if (newBackupEnabled != backupEnabled) {
            backupEnabled = newBackupEnabled;
            if (!backupEnabled) {
                replicaInfoManager.removeListener(flowTable);
                if (backupTask != null) {
                    backupTask.cancel(false);
                    backupTask = null;
                }
            } else {
                replicaInfoManager.addListener(flowTable);
            }
            restartBackupTask = backupEnabled;
        }
        if (newBackupPeriod != backupPeriod) {
            backupPeriod = newBackupPeriod;
            restartBackupTask = backupEnabled;
        }
        if (restartBackupTask) {
            if (backupTask != null) {
                // cancel previously running task
                backupTask.cancel(false);
            }
            backupTask = backupSenderExecutor.scheduleWithFixedDelay(
                    flowTable::backup,
                    0,
                    backupPeriod,
                    TimeUnit.MILLISECONDS);
        }
        if (newPoolSize != msgHandlerPoolSize) {
            msgHandlerPoolSize = newPoolSize;
            ExecutorService oldMsgHandler = messageHandlingExecutor;
            messageHandlingExecutor = Executors.newFixedThreadPool(
                    msgHandlerPoolSize, groupedThreads("onos/store/flow", "message-handlers"));

            // replace previously registered handlers.
            registerMessageHandlers(messageHandlingExecutor);
            oldMsgHandler.shutdown();
        }
        logConfig("Reconfigured");
    }

    private void registerMessageHandlers(ExecutorService executor) {

        clusterCommunicator.addSubscriber(APPLY_BATCH_FLOWS, new OnStoreBatch(), executor);
        clusterCommunicator.<FlowRuleBatchEvent>addSubscriber(
                REMOTE_APPLY_COMPLETED, SERIALIZER::decode, this::notifyDelegate, executor);
        clusterCommunicator.addSubscriber(
                GET_FLOW_ENTRY, SERIALIZER::decode, flowTable::getFlowEntry, SERIALIZER::encode, executor);
        clusterCommunicator.addSubscriber(
                GET_DEVICE_FLOW_ENTRIES, SERIALIZER::decode, flowTable::getFlowEntries, SERIALIZER::encode, executor);
        clusterCommunicator.addSubscriber(
                REMOVE_FLOW_ENTRY, SERIALIZER::decode, this::removeFlowRuleInternal, SERIALIZER::encode, executor);
        clusterCommunicator.addSubscriber(
                REMOVE_FLOW_ENTRY, SERIALIZER::decode, this::removeFlowRuleInternal, SERIALIZER::encode, executor);
        clusterCommunicator.addSubscriber(
                FLOW_TABLE_BACKUP, SERIALIZER::decode, flowTable::onBackupReceipt, SERIALIZER::encode, executor);
    }

    private void unregisterMessageHandlers() {
        clusterCommunicator.removeSubscriber(REMOVE_FLOW_ENTRY);
        clusterCommunicator.removeSubscriber(GET_DEVICE_FLOW_ENTRIES);
        clusterCommunicator.removeSubscriber(GET_FLOW_ENTRY);
        clusterCommunicator.removeSubscriber(APPLY_BATCH_FLOWS);
        clusterCommunicator.removeSubscriber(REMOTE_APPLY_COMPLETED);
        clusterCommunicator.removeSubscriber(FLOW_TABLE_BACKUP);
    }

    private void logConfig(String prefix) {
        log.info("{} with msgHandlerPoolSize = {}; backupEnabled = {}, backupPeriod = {}",
                 prefix, msgHandlerPoolSize, backupEnabled, backupPeriod);
    }

    // This is not a efficient operation on a distributed sharded
    // flow store. We need to revisit the need for this operation or at least
    // make it device specific.
    @Override
    public int getFlowRuleCount() {
        AtomicInteger sum = new AtomicInteger(0);
        deviceService.getDevices().forEach(device -> sum.addAndGet(Iterables.size(getFlowEntries(device.id()))));
        return sum.get();
    }

    @Override
    public FlowEntry getFlowEntry(FlowRule rule) {
        NodeId master = mastershipService.getMasterFor(rule.deviceId());

        if (master == null) {
            log.debug("Failed to getFlowEntry: No master for {}", rule.deviceId());
            return null;
        }

        if (Objects.equal(local, master)) {
            return flowTable.getFlowEntry(rule);
        }

        log.trace("Forwarding getFlowEntry to {}, which is the primary (master) for device {}",
                  master, rule.deviceId());

        return Tools.futureGetOrElse(clusterCommunicator.sendAndReceive(rule,
                                    FlowStoreMessageSubjects.GET_FLOW_ENTRY,
                                    SERIALIZER::encode,
                                    SERIALIZER::decode,
                                    master),
                               FLOW_RULE_STORE_TIMEOUT_MILLIS,
                               TimeUnit.MILLISECONDS,
                               null);
    }

    @Override
    public Iterable<FlowEntry> getFlowEntries(DeviceId deviceId) {
        NodeId master = mastershipService.getMasterFor(deviceId);

        if (master == null) {
            log.debug("Failed to getFlowEntries: No master for {}", deviceId);
            return Collections.emptyList();
        }

        if (Objects.equal(local, master)) {
            return flowTable.getFlowEntries(deviceId);
        }

        log.trace("Forwarding getFlowEntries to {}, which is the primary (master) for device {}",
                  master, deviceId);

        return Tools.futureGetOrElse(clusterCommunicator.sendAndReceive(deviceId,
                                    FlowStoreMessageSubjects.GET_DEVICE_FLOW_ENTRIES,
                                    SERIALIZER::encode,
                                    SERIALIZER::decode,
                                    master),
                               FLOW_RULE_STORE_TIMEOUT_MILLIS,
                               TimeUnit.MILLISECONDS,
                               Collections.emptyList());
    }

    @Override
    public void storeFlowRule(FlowRule rule) {
        storeBatch(new FlowRuleBatchOperation(
                Collections.singletonList(new FlowRuleBatchEntry(FlowRuleOperation.ADD, rule)),
                rule.deviceId(), idGenerator.getNewId()));
    }

    @Override
    public void storeBatch(FlowRuleBatchOperation operation) {
        if (operation.getOperations().isEmpty()) {
            notifyDelegate(FlowRuleBatchEvent.completed(
                    new FlowRuleBatchRequest(operation.id(), Collections.emptySet()),
                    new CompletedBatchOperation(true, Collections.emptySet(), operation.deviceId())));
            return;
        }

        DeviceId deviceId = operation.deviceId();
        NodeId master = mastershipService.getMasterFor(deviceId);

        if (master == null) {
            log.warn("No master for {} : flows will be marked for removal", deviceId);

            updateStoreInternal(operation);

            notifyDelegate(FlowRuleBatchEvent.completed(
                    new FlowRuleBatchRequest(operation.id(), Collections.emptySet()),
                    new CompletedBatchOperation(true, Collections.emptySet(), operation.deviceId())));
            return;
        }

        if (Objects.equal(local, master)) {
            storeBatchInternal(operation);
            return;
        }

        log.trace("Forwarding storeBatch to {}, which is the primary (master) for device {}",
                  master, deviceId);

        clusterCommunicator.unicast(operation,
                                    APPLY_BATCH_FLOWS,
                                    SERIALIZER::encode,
                                    master)
                           .whenComplete((result, error) -> {
                               if (error != null) {
                                   log.warn("Failed to storeBatch: {} to {}", operation, master, error);

                                   Set<FlowRule> allFailures = operation.getOperations()
                                           .stream()
                                           .map(op -> op.target())
                                           .collect(Collectors.toSet());

                                   notifyDelegate(FlowRuleBatchEvent.completed(
                                           new FlowRuleBatchRequest(operation.id(), Collections.emptySet()),
                                           new CompletedBatchOperation(false, allFailures, deviceId)));
                               }
                           });
    }

    private void storeBatchInternal(FlowRuleBatchOperation operation) {

        final DeviceId did = operation.deviceId();
        //final Collection<FlowEntry> ft = flowTable.getFlowEntries(did);
        Set<FlowRuleBatchEntry> currentOps = updateStoreInternal(operation);
        if (currentOps.isEmpty()) {
            batchOperationComplete(FlowRuleBatchEvent.completed(
                    new FlowRuleBatchRequest(operation.id(), Collections.emptySet()),
                    new CompletedBatchOperation(true, Collections.emptySet(), did)));
            return;
        }

        notifyDelegate(FlowRuleBatchEvent.requested(new
                           FlowRuleBatchRequest(operation.id(),
                                                currentOps), operation.deviceId()));
    }

    private Set<FlowRuleBatchEntry> updateStoreInternal(FlowRuleBatchOperation operation) {
        return operation.getOperations().stream().map(
                op -> {
                    StoredFlowEntry entry;
                    switch (op.operator()) {
                        case ADD:
                            entry = new DefaultFlowEntry(op.target());
                            // always add requested FlowRule
                            // Note: 2 equal FlowEntry may have different treatment
                            flowTable.remove(entry.deviceId(), entry);
                            flowTable.add(entry);

                            return op;
                        case REMOVE:
                            entry = flowTable.getFlowEntry(op.target());
                            if (entry != null) {
                                entry.setState(FlowEntryState.PENDING_REMOVE);
                                return op;
                            }
                            break;
                        case MODIFY:
                            //TODO: figure this out at some point
                            break;
                        default:
                            log.warn("Unknown flow operation operator: {}", op.operator());
                    }
                    return null;
                }
        ).filter(op -> op != null).collect(Collectors.toSet());
    }

    @Override
    public void deleteFlowRule(FlowRule rule) {
        storeBatch(
                new FlowRuleBatchOperation(
                        Collections.singletonList(
                                new FlowRuleBatchEntry(
                                        FlowRuleOperation.REMOVE,
                                        rule)), rule.deviceId(), idGenerator.getNewId()));
    }

    @Override
    public FlowRuleEvent addOrUpdateFlowRule(FlowEntry rule) {
        NodeId master = mastershipService.getMasterFor(rule.deviceId());
        if (Objects.equal(local, master)) {
            return addOrUpdateFlowRuleInternal(rule);
        }

        log.warn("Tried to update FlowRule {} state,"
                         + " while the Node was not the master.", rule);
        return null;
    }

    private FlowRuleEvent addOrUpdateFlowRuleInternal(FlowEntry rule) {
        // check if this new rule is an update to an existing entry
        StoredFlowEntry stored = flowTable.getFlowEntry(rule);
        if (stored != null) {
            stored.setBytes(rule.bytes());
            stored.setLife(rule.life());
            stored.setPackets(rule.packets());
            if (stored.state() == FlowEntryState.PENDING_ADD) {
                stored.setState(FlowEntryState.ADDED);
                return new FlowRuleEvent(Type.RULE_ADDED, rule);
            }
            return new FlowRuleEvent(Type.RULE_UPDATED, rule);
        }

        // TODO: Confirm if this behavior is correct. See SimpleFlowRuleStore
        // TODO: also update backup if the behavior is correct.
        flowTable.add(rule);
        return null;
    }

    @Override
    public FlowRuleEvent removeFlowRule(FlowEntry rule) {
        final DeviceId deviceId = rule.deviceId();
        NodeId master = mastershipService.getMasterFor(deviceId);

        if (Objects.equal(local, master)) {
            // bypass and handle it locally
            return removeFlowRuleInternal(rule);
        }

        if (master == null) {
            log.warn("Failed to removeFlowRule: No master for {}", deviceId);
            // TODO: revisit if this should be null (="no-op") or Exception
            return null;
        }

        log.trace("Forwarding removeFlowRule to {}, which is the master for device {}",
                  master, deviceId);

        return Futures.get(clusterCommunicator.sendAndReceive(
                               rule,
                               REMOVE_FLOW_ENTRY,
                               SERIALIZER::encode,
                               SERIALIZER::decode,
                               master),
                           FLOW_RULE_STORE_TIMEOUT_MILLIS,
                           TimeUnit.MILLISECONDS,
                           RuntimeException.class);
    }

    private FlowRuleEvent removeFlowRuleInternal(FlowEntry rule) {
        final DeviceId deviceId = rule.deviceId();
        // This is where one could mark a rule as removed and still keep it in the store.
        final boolean removed = flowTable.remove(deviceId, rule); //flowEntries.remove(deviceId, rule);
        return removed ? new FlowRuleEvent(RULE_REMOVED, rule) : null;
    }

    @Override
    public void batchOperationComplete(FlowRuleBatchEvent event) {
        //FIXME: need a per device pending response
        NodeId nodeId = pendingResponses.remove(event.subject().batchId());
        if (nodeId == null) {
            notifyDelegate(event);
        } else {
            // TODO check unicast return value
            clusterCommunicator.unicast(event, REMOTE_APPLY_COMPLETED, SERIALIZER::encode, nodeId);
            //error log: log.warn("Failed to respond to peer for batch operation result");
        }
    }

    private final class OnStoreBatch implements ClusterMessageHandler {

        @Override
        public void handle(final ClusterMessage message) {
            FlowRuleBatchOperation operation = SERIALIZER.decode(message.payload());
            log.debug("received batch request {}", operation);

            final DeviceId deviceId = operation.deviceId();
            NodeId master = mastershipService.getMasterFor(deviceId);
            if (!Objects.equal(local, master)) {
                Set<FlowRule> failures = new HashSet<>(operation.size());
                for (FlowRuleBatchEntry op : operation.getOperations()) {
                    failures.add(op.target());
                }
                CompletedBatchOperation allFailed = new CompletedBatchOperation(false, failures, deviceId);
                // This node is no longer the master, respond as all failed.
                // TODO: we might want to wrap response in envelope
                // to distinguish sw programming failure and hand over
                // it make sense in the latter case to retry immediately.
                message.respond(SERIALIZER.encode(allFailed));
                return;
            }

            pendingResponses.put(operation.id(), message.sender());
            storeBatchInternal(operation);
        }
    }

    private class InternalFlowTable implements ReplicaInfoEventListener {

        private final Map<DeviceId, Map<FlowId, Set<StoredFlowEntry>>>
                flowEntries = Maps.newConcurrentMap();

        private final Map<DeviceId, Long> lastBackupTimes = Maps.newConcurrentMap();
        private final Map<DeviceId, Long> lastUpdateTimes = Maps.newConcurrentMap();
        private final Map<DeviceId, NodeId> lastBackupNodes = Maps.newConcurrentMap();

        @Override
        public void event(ReplicaInfoEvent event) {
            if (!backupEnabled) {
                return;
            }
            if (event.type() == ReplicaInfoEvent.Type.BACKUPS_CHANGED) {
                DeviceId deviceId = event.subject();
                NodeId master = mastershipService.getMasterFor(deviceId);
                if (!Objects.equal(local, master)) {
                 // ignore since this event is for a device this node does not manage.
                    return;
                }
                NodeId newBackupNode = getBackupNode(deviceId);
                NodeId currentBackupNode = lastBackupNodes.get(deviceId);
                if (Objects.equal(newBackupNode, currentBackupNode)) {
                    // ignore since backup location hasn't changed.
                    return;
                }
                if (currentBackupNode != null && newBackupNode == null) {
                    // Current backup node is most likely down and no alternate backup node
                    // has been chosen. Clear current backup location so that we can resume
                    // backups when either current backup comes online or a different backup node
                    // is chosen.
                    log.warn("Lost backup location {} for deviceId {} and no alternate backup node exists. "
                            + "Flows can be lost if the master goes down", currentBackupNode, deviceId);
                    lastBackupNodes.remove(deviceId);
                    lastBackupTimes.remove(deviceId);
                    return;
                    // TODO: Pick any available node as backup and ensure hand-off occurs when
                    // a new master is elected.
                }
                log.debug("Backup location for {} has changed from {} to {}.",
                        deviceId, currentBackupNode, newBackupNode);
                backupSenderExecutor.schedule(() -> backupFlowEntries(newBackupNode, Sets.newHashSet(deviceId)),
                        0,
                        TimeUnit.SECONDS);
            }
        }

        private void sendBackups(NodeId nodeId, Set<DeviceId> deviceIds) {
            // split up the devices into smaller batches and send them separately.
            Iterables.partition(deviceIds, FLOW_TABLE_BACKUP_BATCH_SIZE)
                     .forEach(ids -> backupFlowEntries(nodeId, Sets.newHashSet(ids)));
        }

        private void backupFlowEntries(NodeId nodeId, Set<DeviceId> deviceIds) {
            if (deviceIds.isEmpty()) {
                return;
            }
            log.debug("Sending flowEntries for devices {} to {} as backup.", deviceIds, nodeId);
            Map<DeviceId, Map<FlowId, Set<StoredFlowEntry>>> deviceFlowEntries =
                    Maps.newConcurrentMap();
            deviceIds.forEach(id -> deviceFlowEntries.put(id, ImmutableMap.copyOf(getFlowTable(id))));
            clusterCommunicator.<Map<DeviceId, Map<FlowId, Set<StoredFlowEntry>>>, Set<DeviceId>>sendAndReceive(
                                        deviceFlowEntries,
                                        FLOW_TABLE_BACKUP,
                                        SERIALIZER::encode,
                                        SERIALIZER::decode,
                                        nodeId)
                               .whenComplete((backedupDevices, error) -> {
                                   Set<DeviceId> devicesNotBackedup = error != null ?
                                           deviceFlowEntries.keySet() :
                                           Sets.difference(deviceFlowEntries.keySet(), backedupDevices);
                                   if (devicesNotBackedup.size() > 0) {
                                       log.warn("Failed to backup devices: {}. Reason: {}",
                                               devicesNotBackedup, error.getMessage());
                                   }
                                   if (backedupDevices != null) {
                                       backedupDevices.forEach(id -> {
                                           lastBackupTimes.put(id, System.currentTimeMillis());
                                           lastBackupNodes.put(id, nodeId);
                                       });
                                   }
                               });
        }

        /**
         * Returns the flow table for specified device.
         *
         * @param deviceId identifier of the device
         * @return Map representing Flow Table of given device.
         */
        private Map<FlowId, Set<StoredFlowEntry>> getFlowTable(DeviceId deviceId) {
            if (persistenceEnabled) {
                return flowEntries.computeIfAbsent(deviceId, id -> persistenceService
                        .<FlowId, Set<StoredFlowEntry>>persistentMapBuilder()
                        .withName("FlowTable:" + deviceId.toString())
                        .withSerializer(new Serializer() {
                            @Override
                            public <T> byte[] encode(T object) {
                                return SERIALIZER.encode(object);
                            }

                            @Override
                            public <T> T decode(byte[] bytes) {
                                return SERIALIZER.decode(bytes);
                            }
                        })
                        .build());
            } else {
                return flowEntries.computeIfAbsent(deviceId, id -> Maps.newConcurrentMap());
            }
        }

        private Set<StoredFlowEntry> getFlowEntriesInternal(DeviceId deviceId, FlowId flowId) {
            return getFlowTable(deviceId).computeIfAbsent(flowId, id -> Sets.newCopyOnWriteArraySet());
        }

        private StoredFlowEntry getFlowEntryInternal(FlowRule rule) {
            Set<StoredFlowEntry> flowEntries = getFlowEntriesInternal(rule.deviceId(), rule.id());
            return flowEntries.stream()
                              .filter(entry -> Objects.equal(entry, rule))
                              .findAny()
                              .orElse(null);
        }

        private Set<FlowEntry> getFlowEntriesInternal(DeviceId deviceId) {
            Set<FlowEntry> result = Sets.newHashSet();
            getFlowTable(deviceId).values().forEach(result::addAll);
            return result;
        }

        public StoredFlowEntry getFlowEntry(FlowRule rule) {
            return getFlowEntryInternal(rule);
        }

        public Set<FlowEntry> getFlowEntries(DeviceId deviceId) {
            return getFlowEntriesInternal(deviceId);
        }

        public void add(FlowEntry rule) {
            getFlowEntriesInternal(rule.deviceId(), rule.id()).add((StoredFlowEntry) rule);
            lastUpdateTimes.put(rule.deviceId(), System.currentTimeMillis());
        }

        public boolean remove(DeviceId deviceId, FlowEntry rule) {
            try {
                return getFlowEntriesInternal(deviceId, rule.id()).remove(rule);
            } finally {
                lastUpdateTimes.put(deviceId, System.currentTimeMillis());
            }
        }

        private NodeId getBackupNode(DeviceId deviceId) {
            List<NodeId> deviceStandbys = replicaInfoManager.getReplicaInfoFor(deviceId).backups();
            // pick the standby which is most likely to become next master
            return deviceStandbys.isEmpty() ? null : deviceStandbys.get(0);
        }

        private void backup() {
            if (!backupEnabled) {
                return;
            }
            try {
                // determine the set of devices that we need to backup during this run.
                Set<DeviceId> devicesToBackup = mastershipService.getDevicesOf(local)
                            .stream()
                            .filter(deviceId -> {
                                Long lastBackupTime = lastBackupTimes.get(deviceId);
                                Long lastUpdateTime = lastUpdateTimes.get(deviceId);
                                NodeId lastBackupNode = lastBackupNodes.get(deviceId);
                                NodeId newBackupNode = getBackupNode(deviceId);
                                return lastBackupTime == null
                                        ||  !Objects.equal(lastBackupNode, newBackupNode)
                                        || (lastUpdateTime != null && lastUpdateTime > lastBackupTime);
                            })
                            .collect(Collectors.toSet());

                // compute a mapping from node to the set of devices whose flow entries it should backup
                Map<NodeId, Set<DeviceId>> devicesToBackupByNode = Maps.newHashMap();
                devicesToBackup.forEach(deviceId -> {
                    NodeId backupLocation = getBackupNode(deviceId);
                    if (backupLocation != null) {
                        devicesToBackupByNode.computeIfAbsent(backupLocation, nodeId -> Sets.newHashSet())
                                             .add(deviceId);
                    }
                });
                // send the device flow entries to their respective backup nodes
                devicesToBackupByNode.forEach(this::sendBackups);
            } catch (Exception e) {
                log.error("Backup failed.", e);
            }
        }

        private Set<DeviceId> onBackupReceipt(Map<DeviceId, Map<FlowId, Set<StoredFlowEntry>>> flowTables) {
            log.debug("Received flowEntries for {} to backup", flowTables.keySet());
            Set<DeviceId> backedupDevices = Sets.newHashSet();
            try {
                flowTables.forEach((deviceId, deviceFlowTable) -> {
                    // Only process those devices are that not managed by the local node.
                    if (!Objects.equal(local, mastershipService.getMasterFor(deviceId))) {
                        Map<FlowId, Set<StoredFlowEntry>> backupFlowTable = getFlowTable(deviceId);
                        backupFlowTable.clear();
                        backupFlowTable.putAll(deviceFlowTable);
                        backedupDevices.add(deviceId);
                    }
                });
            } catch (Exception e) {
                log.warn("Failure processing backup request", e);
            }
            return backedupDevices;
        }
    }

    @Override
    public FlowRuleEvent updateTableStatistics(DeviceId deviceId,
                                               List<TableStatisticsEntry> tableStats) {
        deviceTableStats.put(deviceId, tableStats);
        return null;
    }

    @Override
    public Iterable<TableStatisticsEntry> getTableStatistics(DeviceId deviceId) {
        NodeId master = mastershipService.getMasterFor(deviceId);

        if (master == null) {
            log.debug("Failed to getTableStats: No master for {}", deviceId);
            return Collections.emptyList();
        }

        List<TableStatisticsEntry> tableStats = deviceTableStats.get(deviceId);
        if (tableStats == null) {
            return Collections.emptyList();
        }
        return ImmutableList.copyOf(tableStats);
    }

    private class InternalTableStatsListener
        implements EventuallyConsistentMapListener<DeviceId, List<TableStatisticsEntry>> {
        @Override
        public void event(EventuallyConsistentMapEvent<DeviceId,
                          List<TableStatisticsEntry>> event) {
            //TODO: Generate an event to listeners (do we need?)
        }
    }
}