opendaylight/md-sal/sal-akka-raft/src/main/java/org/opendaylight/controller/cluster/raft/behaviors/AbstractLeader.java

   1 /*
   2  * Copyright (c) 2014 Cisco Systems, Inc. and others.  All rights reserved.
   3  *
   4  * This program and the accompanying materials are made available under the
   5  * terms of the Eclipse Public License v1.0 which accompanies this distribution,
   6  * and is available at http://www.eclipse.org/legal/epl-v10.html
   7  */
   8
   9 package org.opendaylight.controller.cluster.raft.behaviors;
  10
  11 import akka.actor.ActorRef;
  12 import akka.actor.ActorSelection;
  13 import akka.actor.Cancellable;
  14 import com.google.common.annotations.VisibleForTesting;
  15 import com.google.common.base.Optional;
  16 import com.google.common.base.Preconditions;
  17 import com.google.common.collect.ImmutableMap;
  18 import com.google.common.collect.ImmutableMap.Builder;
  19 import com.google.protobuf.ByteString;
  20 import java.io.IOException;
  21 import java.util.Collection;
  22 import java.util.Collections;
  23 import java.util.HashMap;
  24 import java.util.Iterator;
  25 import java.util.LinkedList;
  26 import java.util.List;
  27 import java.util.Map;
  28 import java.util.Map.Entry;
  29 import java.util.concurrent.TimeUnit;
  30 import org.opendaylight.controller.cluster.raft.ClientRequestTracker;
  31 import org.opendaylight.controller.cluster.raft.ClientRequestTrackerImpl;
  32 import org.opendaylight.controller.cluster.raft.FollowerLogInformation;
  33 import org.opendaylight.controller.cluster.raft.FollowerLogInformationImpl;
  34 import org.opendaylight.controller.cluster.raft.RaftActorContext;
  35 import org.opendaylight.controller.cluster.raft.RaftState;
  36 import org.opendaylight.controller.cluster.raft.ReplicatedLogEntry;
  37 import org.opendaylight.controller.cluster.raft.base.messages.CaptureSnapshot;
  38 import org.opendaylight.controller.cluster.raft.base.messages.InitiateInstallSnapshot;
  39 import org.opendaylight.controller.cluster.raft.base.messages.Replicate;
  40 import org.opendaylight.controller.cluster.raft.base.messages.SendHeartBeat;
  41 import org.opendaylight.controller.cluster.raft.base.messages.SendInstallSnapshot;
  42 import org.opendaylight.controller.cluster.raft.messages.AppendEntries;
  43 import org.opendaylight.controller.cluster.raft.messages.AppendEntriesReply;
  44 import org.opendaylight.controller.cluster.raft.messages.InstallSnapshot;
  45 import org.opendaylight.controller.cluster.raft.messages.InstallSnapshotReply;
  46 import org.opendaylight.controller.cluster.raft.messages.RaftRPC;
  47 import org.opendaylight.controller.cluster.raft.messages.RequestVoteReply;
  48 import scala.concurrent.duration.FiniteDuration;
  49
  50 /**
  51  * The behavior of a RaftActor when it is in the Leader state
  52  * <p/>
  53  * Leaders:
  54  * <ul>
  55  * <li> Upon election: send initial empty AppendEntries RPCs
  56  * (heartbeat) to each server; repeat during idle periods to
  57  * prevent election timeouts (§5.2)
  58  * <li> If command received from client: append entry to local log,
  59  * respond after entry applied to state machine (§5.3)
  60  * <li> If last log index ≥ nextIndex for a follower: send
  61  * AppendEntries RPC with log entries starting at nextIndex
  62  * <ul>
  63  * <li> If successful: update nextIndex and matchIndex for
  64  * follower (§5.3)
  65  * <li> If AppendEntries fails because of log inconsistency:
  66  * decrement nextIndex and retry (§5.3)
  67  * </ul>
  68  * <li> If there exists an N such that N > commitIndex, a majority
  69  * of matchIndex[i] ≥ N, and log[N].term == currentTerm:
  70  * set commitIndex = N (§5.3, §5.4).
  71  */
  72 public abstract class AbstractLeader extends AbstractRaftActorBehavior {
  73
  74     // The index of the first chunk that is sent when installing a snapshot
  75     public static final int FIRST_CHUNK_INDEX = 1;
  76
  77     // The index that the follower should respond with if it needs the install snapshot to be reset
  78     public static final int INVALID_CHUNK_INDEX = -1;
  79
  80     // This would be passed as the hash code of the last chunk when sending the first chunk
  81     public static final int INITIAL_LAST_CHUNK_HASH_CODE = -1;
  82
  83     private final Map<String, FollowerLogInformation> followerToLog;
  84     private final Map<String, FollowerToSnapshot> mapFollowerToSnapshot = new HashMap<>();
  85
  86     private Cancellable heartbeatSchedule = null;
  87
  88     private final Collection<ClientRequestTracker> trackerList = new LinkedList<>();
  89
  90     protected final int minReplicationCount;
  91
  92     protected final int minIsolatedLeaderPeerCount;
  93
  94     private Optional<ByteString> snapshot;
  95
  96     private long replicatedToAllIndex = -1;
  97
  98     public AbstractLeader(RaftActorContext context) {
  99         super(context);
 100
 101         final Builder<String, FollowerLogInformation> ftlBuilder = ImmutableMap.builder();
 102         for (String followerId : context.getPeerAddresses().keySet()) {
 103             FollowerLogInformation followerLogInformation =
 104                 new FollowerLogInformationImpl(followerId,
 105                     context.getCommitIndex(), -1,
 106                     context.getConfigParams().getElectionTimeOutInterval());
 107
 108             ftlBuilder.put(followerId, followerLogInformation);
 109         }
 110         followerToLog = ftlBuilder.build();
 111
 112         leaderId = context.getId();
 113
 114         LOG.debug("{}: Election: Leader has following peers: {}", context.getId(), getFollowerIds());
 115
 116         minReplicationCount = getMajorityVoteCount(getFollowerIds().size());
 117
 118         // the isolated Leader peer count will be 1 less than the majority vote count.
 119         // this is because the vote count has the self vote counted in it
 120         // for e.g
 121         // 0 peers = 1 votesRequired , minIsolatedLeaderPeerCount = 0
 122         // 2 peers = 2 votesRequired , minIsolatedLeaderPeerCount = 1
 123         // 4 peers = 3 votesRequired, minIsolatedLeaderPeerCount = 2
 124         minIsolatedLeaderPeerCount = minReplicationCount > 0 ? (minReplicationCount - 1) : 0;
 125
 126         snapshot = Optional.absent();
 127
 128         // Immediately schedule a heartbeat
 129         // Upon election: send initial empty AppendEntries RPCs
 130         // (heartbeat) to each server; repeat during idle periods to
 131         // prevent election timeouts (§5.2)
 132         scheduleHeartBeat(new FiniteDuration(0, TimeUnit.SECONDS));
 133     }
 134
 135     /**
 136      * Return an immutable collection of follower identifiers.
 137      *
 138      * @return Collection of follower IDs
 139      */
 140     protected final Collection<String> getFollowerIds() {
 141         return followerToLog.keySet();
 142     }
 143
 144     private Optional<ByteString> getSnapshot() {
 145         return snapshot;
 146     }
 147
 148     @VisibleForTesting
 149     void setSnapshot(Optional<ByteString> snapshot) {
 150         this.snapshot = snapshot;
 151     }
 152
 153     @Override
 154     protected RaftActorBehavior handleAppendEntries(ActorRef sender,
 155         AppendEntries appendEntries) {
 156
 157         if(LOG.isDebugEnabled()) {
 158             LOG.debug("{}: handleAppendEntries: {}", context.getId(), appendEntries);
 159         }
 160
 161         return this;
 162     }
 163
 164     @Override
 165     protected RaftActorBehavior handleAppendEntriesReply(ActorRef sender,
 166         AppendEntriesReply appendEntriesReply) {
 167
 168         if(! appendEntriesReply.isSuccess()) {
 169             if(LOG.isDebugEnabled()) {
 170                 LOG.debug("{}: handleAppendEntriesReply: {}", context.getId(), appendEntriesReply);
 171             }
 172         }
 173
 174         // Update the FollowerLogInformation
 175         String followerId = appendEntriesReply.getFollowerId();
 176         FollowerLogInformation followerLogInformation =
 177             followerToLog.get(followerId);
 178
 179         if(followerLogInformation == null){
 180             LOG.error("{}: handleAppendEntriesReply - unknown follower {}", context.getId(), followerId);
 181             return this;
 182         }
 183
 184         followerLogInformation.markFollowerActive();
 185
 186         if (appendEntriesReply.isSuccess()) {
 187             followerLogInformation
 188                 .setMatchIndex(appendEntriesReply.getLogLastIndex());
 189             followerLogInformation
 190                 .setNextIndex(appendEntriesReply.getLogLastIndex() + 1);
 191         } else {
 192
 193             // TODO: When we find that the follower is out of sync with the
 194             // Leader we simply decrement that followers next index by 1.
 195             // Would it be possible to do better than this? The RAFT spec
 196             // does not explicitly deal with it but may be something for us to
 197             // think about
 198
 199             followerLogInformation.decrNextIndex();
 200         }
 201
 202         // Now figure out if this reply warrants a change in the commitIndex
 203         // If there exists an N such that N > commitIndex, a majority
 204         // of matchIndex[i] ≥ N, and log[N].term == currentTerm:
 205         // set commitIndex = N (§5.3, §5.4).
 206         for (long N = context.getCommitIndex() + 1; ; N++) {
 207             int replicatedCount = 1;
 208
 209             for (FollowerLogInformation info : followerToLog.values()) {
 210                 if (info.getMatchIndex() >= N) {
 211                     replicatedCount++;
 212                 }
 213             }
 214
 215             if (replicatedCount >= minReplicationCount) {
 216                 ReplicatedLogEntry replicatedLogEntry = context.getReplicatedLog().get(N);
 217                 if (replicatedLogEntry != null &&
 218                     replicatedLogEntry.getTerm() == currentTerm()) {
 219                     context.setCommitIndex(N);
 220                 }
 221             } else {
 222                 break;
 223             }
 224         }
 225
 226         // Apply the change to the state machine
 227         if (context.getCommitIndex() > context.getLastApplied()) {
 228             applyLogToStateMachine(context.getCommitIndex());
 229         }
 230
 231         if (!context.isSnapshotCaptureInitiated()) {
 232             purgeInMemoryLog();
 233         }
 234
 235         return this;
 236     }
 237
 238     private void purgeInMemoryLog() {
 239         //find the lowest index across followers which has been replicated to all. -1 if there are no followers.
 240         // we would delete the in-mem log from that index on, in-order to minimize mem usage
 241         // we would also share this info thru AE with the followers so that they can delete their log entries as well.
 242         long minReplicatedToAllIndex = followerToLog.isEmpty() ? -1 : Long.MAX_VALUE;
 243         for (FollowerLogInformation info : followerToLog.values()) {
 244             minReplicatedToAllIndex = Math.min(minReplicatedToAllIndex, info.getMatchIndex());
 245         }
 246
 247         replicatedToAllIndex = fakeSnapshot(minReplicatedToAllIndex, replicatedToAllIndex);
 248     }
 249
 250     @Override
 251     protected ClientRequestTracker removeClientRequestTracker(long logIndex) {
 252         final Iterator<ClientRequestTracker> it = trackerList.iterator();
 253         while (it.hasNext()) {
 254             final ClientRequestTracker t = it.next();
 255             if (t.getIndex() == logIndex) {
 256                 it.remove();
 257                 return t;
 258             }
 259         }
 260
 261         return null;
 262     }
 263
 264     @Override
 265     protected ClientRequestTracker findClientRequestTracker(long logIndex) {
 266         for (ClientRequestTracker tracker : trackerList) {
 267             if (tracker.getIndex() == logIndex) {
 268                 return tracker;
 269             }
 270         }
 271         return null;
 272     }
 273
 274     @Override
 275     protected RaftActorBehavior handleRequestVoteReply(ActorRef sender,
 276         RequestVoteReply requestVoteReply) {
 277         return this;
 278     }
 279
 280     @Override
 281     public RaftState state() {
 282         return RaftState.Leader;
 283     }
 284
 285     @Override
 286     public RaftActorBehavior handleMessage(ActorRef sender, Object originalMessage) {
 287         Preconditions.checkNotNull(sender, "sender should not be null");
 288
 289         Object message = fromSerializableMessage(originalMessage);
 290
 291         if (message instanceof RaftRPC) {
 292             RaftRPC rpc = (RaftRPC) message;
 293             // If RPC request or response contains term T > currentTerm:
 294             // set currentTerm = T, convert to follower (§5.1)
 295             // This applies to all RPC messages and responses
 296             if (rpc.getTerm() > context.getTermInformation().getCurrentTerm()) {
 297                 LOG.debug("{}: Term {} in \"{}\" message is greater than leader's term {}", context.getId(),
 298                         rpc.getTerm(), rpc, context.getTermInformation().getCurrentTerm());
 299
 300                 context.getTermInformation().updateAndPersist(rpc.getTerm(), null);
 301
 302                 return switchBehavior(new Follower(context));
 303             }
 304         }
 305
 306         try {
 307             if (message instanceof SendHeartBeat) {
 308                 sendHeartBeat();
 309                 return this;
 310
 311             } else if(message instanceof InitiateInstallSnapshot) {
 312                 installSnapshotIfNeeded();
 313
 314             } else if(message instanceof SendInstallSnapshot) {
 315                 // received from RaftActor
 316                 setSnapshot(Optional.of(((SendInstallSnapshot) message).getSnapshot()));
 317                 sendInstallSnapshot();
 318
 319             } else if (message instanceof Replicate) {
 320                 replicate((Replicate) message);
 321
 322             } else if (message instanceof InstallSnapshotReply){
 323                 handleInstallSnapshotReply((InstallSnapshotReply) message);
 324
 325             }
 326         } finally {
 327             scheduleHeartBeat(context.getConfigParams().getHeartBeatInterval());
 328         }
 329
 330         return super.handleMessage(sender, message);
 331     }
 332
 333     private void handleInstallSnapshotReply(InstallSnapshotReply reply) {
 334         String followerId = reply.getFollowerId();
 335         FollowerToSnapshot followerToSnapshot = mapFollowerToSnapshot.get(followerId);
 336
 337         if (followerToSnapshot == null) {
 338             LOG.error("{}: FollowerId {} in InstallSnapshotReply not known to Leader",
 339                     context.getId(), followerId);
 340             return;
 341         }
 342
 343         FollowerLogInformation followerLogInformation = followerToLog.get(followerId);
 344         followerLogInformation.markFollowerActive();
 345
 346         if (followerToSnapshot.getChunkIndex() == reply.getChunkIndex()) {
 347             if (reply.isSuccess()) {
 348                 if(followerToSnapshot.isLastChunk(reply.getChunkIndex())) {
 349                     //this was the last chunk reply
 350                     if(LOG.isDebugEnabled()) {
 351                         LOG.debug("{}: InstallSnapshotReply received, " +
 352                                 "last chunk received, Chunk:{}. Follower:{} Setting nextIndex:{}",
 353                                 context.getId(), reply.getChunkIndex(), followerId,
 354                             context.getReplicatedLog().getSnapshotIndex() + 1
 355                         );
 356                     }
 357
 358                     followerLogInformation.setMatchIndex(
 359                         context.getReplicatedLog().getSnapshotIndex());
 360                     followerLogInformation.setNextIndex(
 361                         context.getReplicatedLog().getSnapshotIndex() + 1);
 362                     mapFollowerToSnapshot.remove(followerId);
 363
 364                     if(LOG.isDebugEnabled()) {
 365                         LOG.debug("{}: followerToLog.get(followerId).getNextIndex()=" +
 366                                 context.getId(), followerToLog.get(followerId).getNextIndex());
 367                     }
 368
 369                     if (mapFollowerToSnapshot.isEmpty()) {
 370                         // once there are no pending followers receiving snapshots
 371                         // we can remove snapshot from the memory
 372                         setSnapshot(Optional.<ByteString>absent());
 373                     }
 374
 375                 } else {
 376                     followerToSnapshot.markSendStatus(true);
 377                 }
 378             } else {
 379                 LOG.info("{}: InstallSnapshotReply received sending snapshot chunk failed, Will retry, Chunk: {}",
 380                         context.getId(), reply.getChunkIndex());
 381
 382                 followerToSnapshot.markSendStatus(false);
 383             }
 384         } else {
 385             LOG.error("{}: Chunk index {} in InstallSnapshotReply from follower {} does not match expected index {}",
 386                     context.getId(), reply.getChunkIndex(), followerId,
 387                     followerToSnapshot.getChunkIndex());
 388
 389             if(reply.getChunkIndex() == INVALID_CHUNK_INDEX){
 390                 // Since the Follower did not find this index to be valid we should reset the follower snapshot
 391                 // so that Installing the snapshot can resume from the beginning
 392                 followerToSnapshot.reset();
 393             }
 394         }
 395     }
 396
 397     private void replicate(Replicate replicate) {
 398         long logIndex = replicate.getReplicatedLogEntry().getIndex();
 399
 400         if(LOG.isDebugEnabled()) {
 401             LOG.debug("{}: Replicate message {}", context.getId(), logIndex);
 402         }
 403
 404         // Create a tracker entry we will use this later to notify the
 405         // client actor
 406         trackerList.add(
 407             new ClientRequestTrackerImpl(replicate.getClientActor(),
 408                 replicate.getIdentifier(),
 409                 logIndex)
 410         );
 411
 412         if (followerToLog.isEmpty()) {
 413             context.setCommitIndex(logIndex);
 414             applyLogToStateMachine(logIndex);
 415         } else {
 416             sendAppendEntries();
 417         }
 418     }
 419
 420     private void sendAppendEntries() {
 421         // Send an AppendEntries to all followers
 422
 423         for (Entry<String, FollowerLogInformation> e : followerToLog.entrySet()) {
 424             final String followerId = e.getKey();
 425             ActorSelection followerActor = context.getPeerActorSelection(followerId);
 426
 427             if (followerActor != null) {
 428                 FollowerLogInformation followerLogInformation = followerToLog.get(followerId);
 429                 long followerNextIndex = followerLogInformation.getNextIndex();
 430                 boolean isFollowerActive = followerLogInformation.isFollowerActive();
 431
 432                 FollowerToSnapshot followerToSnapshot = mapFollowerToSnapshot.get(followerId);
 433                 if (followerToSnapshot != null) {
 434                     // if install snapshot is in process , then sent next chunk if possible
 435                     if (isFollowerActive && followerToSnapshot.canSendNextChunk()) {
 436                         sendSnapshotChunk(followerActor, followerId);
 437                     } else {
 438                         // we send a heartbeat even if we have not received a reply for the last chunk
 439                         sendAppendEntriesToFollower(followerActor, followerNextIndex,
 440                             Collections.<ReplicatedLogEntry>emptyList(), followerId);
 441                     }
 442
 443                 } else {
 444                     long leaderLastIndex = context.getReplicatedLog().lastIndex();
 445                     long leaderSnapShotIndex = context.getReplicatedLog().getSnapshotIndex();
 446                     final List<ReplicatedLogEntry> entries;
 447
 448                     LOG.debug("{}: Checking sendAppendEntries for {}, leaderLastIndex: {}, leaderSnapShotIndex: {}",
 449                             context.getId(), leaderLastIndex, leaderSnapShotIndex);
 450
 451                     if (isFollowerActive && context.getReplicatedLog().isPresent(followerNextIndex)) {
 452                         LOG.debug("{}: sendAppendEntries: {} is present for {}", context.getId(),
 453                                 followerNextIndex, followerId);
 454
 455                         // FIXME : Sending one entry at a time
 456                         entries = context.getReplicatedLog().getFrom(followerNextIndex, 1);
 457
 458                     } else if (isFollowerActive && followerNextIndex >= 0 &&
 459                         leaderLastIndex >= followerNextIndex ) {
 460                         // if the followers next index is not present in the leaders log, and
 461                         // if the follower is just not starting and if leader's index is more than followers index
 462                         // then snapshot should be sent
 463
 464                         if(LOG.isDebugEnabled()) {
 465                             LOG.debug(String.format("%s: InitiateInstallSnapshot to follower: %s," +
 466                                     "follower-nextIndex: %s, leader-snapshot-index: %s,  " +
 467                                     "leader-last-index: %s", context.getId(), followerId,
 468                                 followerNextIndex, leaderSnapShotIndex, leaderLastIndex));
 469                         }
 470                         actor().tell(new InitiateInstallSnapshot(), actor());
 471
 472                         // we would want to sent AE as the capture snapshot might take time
 473                         entries =  Collections.<ReplicatedLogEntry>emptyList();
 474
 475                     } else {
 476                         //we send an AppendEntries, even if the follower is inactive
 477                         // in-order to update the followers timestamp, in case it becomes active again
 478                         entries =  Collections.<ReplicatedLogEntry>emptyList();
 479                     }
 480
 481                     sendAppendEntriesToFollower(followerActor, followerNextIndex, entries, followerId);
 482                 }
 483             }
 484         }
 485     }
 486
 487     private void sendAppendEntriesToFollower(ActorSelection followerActor, long followerNextIndex,
 488         List<ReplicatedLogEntry> entries, String followerId) {
 489         AppendEntries appendEntries = new AppendEntries(currentTerm(), context.getId(),
 490             prevLogIndex(followerNextIndex),
 491             prevLogTerm(followerNextIndex), entries,
 492             context.getCommitIndex(), replicatedToAllIndex);
 493
 494         if(!entries.isEmpty()) {
 495             LOG.debug("{}: Sending AppendEntries to follower {}: {}", context.getId(), followerId,
 496                     appendEntries);
 497         }
 498
 499         followerActor.tell(appendEntries.toSerializable(), actor());
 500     }
 501
 502     /**
 503      * An installSnapshot is scheduled at a interval that is a multiple of
 504      * a HEARTBEAT_INTERVAL. This is to avoid the need to check for installing
 505      * snapshots at every heartbeat.
 506      *
 507      * Install Snapshot works as follows
 508      * 1. Leader sends a InitiateInstallSnapshot message to self
 509      * 2. Leader then initiates the capture snapshot by sending a CaptureSnapshot message to actor
 510      * 3. RaftActor on receipt of the CaptureSnapshotReply (from Shard), stores the received snapshot in the replicated log
 511      * and makes a call to Leader's handleMessage , with SendInstallSnapshot message.
 512      * 4. Leader , picks the snapshot from im-mem ReplicatedLog and sends it in chunks to the Follower
 513      * 5. On complete, Follower sends back a InstallSnapshotReply.
 514      * 6. On receipt of the InstallSnapshotReply for the last chunk, Leader marks the install complete for that follower
 515      * and replenishes the memory by deleting the snapshot in Replicated log.
 516      *
 517      */
 518     private void installSnapshotIfNeeded() {
 519         if(LOG.isDebugEnabled()) {
 520             LOG.debug("{}: installSnapshotIfNeeded, followers {}", context.getId(), followerToLog.keySet());
 521         }
 522
 523         for (Entry<String, FollowerLogInformation> e : followerToLog.entrySet()) {
 524             final ActorSelection followerActor = context.getPeerActorSelection(e.getKey());
 525
 526             if (followerActor != null) {
 527                 long nextIndex = e.getValue().getNextIndex();
 528
 529                 if (!context.getReplicatedLog().isPresent(nextIndex) &&
 530                         context.getReplicatedLog().isInSnapshot(nextIndex)) {
 531                     LOG.info("{}: {} follower needs a snapshot install", context.getId(), e.getKey());
 532                     if (snapshot.isPresent()) {
 533                         // if a snapshot is present in the memory, most likely another install is in progress
 534                         // no need to capture snapshot
 535                         sendSnapshotChunk(followerActor, e.getKey());
 536
 537                     } else if (!context.isSnapshotCaptureInitiated()) {
 538                         initiateCaptureSnapshot();
 539                         //we just need 1 follower who would need snapshot to be installed.
 540                         // when we have the snapshot captured, we would again check (in SendInstallSnapshot)
 541                         // who needs an install and send to all who need
 542                         break;
 543                     }
 544
 545                 }
 546             }
 547         }
 548     }
 549
 550     // on every install snapshot, we try to capture the snapshot.
 551     // Once a capture is going on, another one issued will get ignored by RaftActor.
 552     private void initiateCaptureSnapshot() {
 553         LOG.info("{}: Initiating Snapshot Capture to Install Snapshot, Leader:{}", context.getId(), getLeaderId());
 554         ReplicatedLogEntry lastAppliedEntry = context.getReplicatedLog().get(context.getLastApplied());
 555         long lastAppliedIndex = -1;
 556         long lastAppliedTerm = -1;
 557
 558         if (lastAppliedEntry != null) {
 559             lastAppliedIndex = lastAppliedEntry.getIndex();
 560             lastAppliedTerm = lastAppliedEntry.getTerm();
 561         } else if (context.getReplicatedLog().getSnapshotIndex() > -1)  {
 562             lastAppliedIndex = context.getReplicatedLog().getSnapshotIndex();
 563             lastAppliedTerm = context.getReplicatedLog().getSnapshotTerm();
 564         }
 565
 566         boolean isInstallSnapshotInitiated = true;
 567         actor().tell(new CaptureSnapshot(lastIndex(), lastTerm(),
 568                 lastAppliedIndex, lastAppliedTerm, isInstallSnapshotInitiated),
 569             actor());
 570         context.setSnapshotCaptureInitiated(true);
 571     }
 572
 573
 574     private void sendInstallSnapshot() {
 575         for (Entry<String, FollowerLogInformation> e : followerToLog.entrySet()) {
 576             ActorSelection followerActor = context.getPeerActorSelection(e.getKey());
 577
 578             if (followerActor != null) {
 579                 long nextIndex = e.getValue().getNextIndex();
 580
 581                 if (!context.getReplicatedLog().isPresent(nextIndex) &&
 582                     context.getReplicatedLog().isInSnapshot(nextIndex)) {
 583                     sendSnapshotChunk(followerActor, e.getKey());
 584                 }
 585             }
 586         }
 587     }
 588
 589     /**
 590      *  Sends a snapshot chunk to a given follower
 591      *  InstallSnapshot should qualify as a heartbeat too.
 592      */
 593     private void sendSnapshotChunk(ActorSelection followerActor, String followerId) {
 594         try {
 595             if (snapshot.isPresent()) {
 596                 ByteString nextSnapshotChunk = getNextSnapshotChunk(followerId,snapshot.get());
 597
 598                 // Note: the previous call to getNextSnapshotChunk has the side-effect of adding
 599                 // followerId to the followerToSnapshot map.
 600                 FollowerToSnapshot followerToSnapshot = mapFollowerToSnapshot.get(followerId);
 601
 602                 followerActor.tell(
 603                     new InstallSnapshot(currentTerm(), context.getId(),
 604                         context.getReplicatedLog().getSnapshotIndex(),
 605                         context.getReplicatedLog().getSnapshotTerm(),
 606                         nextSnapshotChunk,
 607                             followerToSnapshot.incrementChunkIndex(),
 608                             followerToSnapshot.getTotalChunks(),
 609                         Optional.of(followerToSnapshot.getLastChunkHashCode())
 610                     ).toSerializable(),
 611                     actor()
 612                 );
 613                 LOG.info("{}: InstallSnapshot sent to follower {}, Chunk: {}/{}",
 614                         context.getId(), followerActor.path(),
 615                         followerToSnapshot.getChunkIndex(),
 616                         followerToSnapshot.getTotalChunks());
 617             }
 618         } catch (IOException e) {
 619             LOG.error(e, "{}: InstallSnapshot failed for Leader.", context.getId());
 620         }
 621     }
 622
 623     /**
 624      * Acccepts snaphot as ByteString, enters into map for future chunks
 625      * creates and return a ByteString chunk
 626      */
 627     private ByteString getNextSnapshotChunk(String followerId, ByteString snapshotBytes) throws IOException {
 628         FollowerToSnapshot followerToSnapshot = mapFollowerToSnapshot.get(followerId);
 629         if (followerToSnapshot == null) {
 630             followerToSnapshot = new FollowerToSnapshot(snapshotBytes);
 631             mapFollowerToSnapshot.put(followerId, followerToSnapshot);
 632         }
 633         ByteString nextChunk = followerToSnapshot.getNextChunk();
 634         if (LOG.isDebugEnabled()) {
 635             LOG.debug("{}: Leader's snapshot nextChunk size:{}", context.getId(), nextChunk.size());
 636         }
 637         return nextChunk;
 638     }
 639
 640     private void sendHeartBeat() {
 641         if (!followerToLog.isEmpty()) {
 642             sendAppendEntries();
 643         }
 644     }
 645
 646     private void stopHeartBeat() {
 647         if (heartbeatSchedule != null && !heartbeatSchedule.isCancelled()) {
 648             heartbeatSchedule.cancel();
 649         }
 650     }
 651
 652     private void scheduleHeartBeat(FiniteDuration interval) {
 653         if (followerToLog.isEmpty()) {
 654             // Optimization - do not bother scheduling a heartbeat as there are
 655             // no followers
 656             return;
 657         }
 658
 659         stopHeartBeat();
 660
 661         // Schedule a heartbeat. When the scheduler triggers a SendHeartbeat
 662         // message is sent to itself.
 663         // Scheduling the heartbeat only once here because heartbeats do not
 664         // need to be sent if there are other messages being sent to the remote
 665         // actor.
 666         heartbeatSchedule = context.getActorSystem().scheduler().scheduleOnce(
 667             interval, context.getActor(), new SendHeartBeat(),
 668             context.getActorSystem().dispatcher(), context.getActor());
 669     }
 670
 671     @Override
 672     public void close() throws Exception {
 673         stopHeartBeat();
 674     }
 675
 676     @Override
 677     public String getLeaderId() {
 678         return context.getId();
 679     }
 680
 681     protected boolean isLeaderIsolated() {
 682         int minPresent = minIsolatedLeaderPeerCount;
 683         for (FollowerLogInformation followerLogInformation : followerToLog.values()) {
 684             if (followerLogInformation.isFollowerActive()) {
 685                 --minPresent;
 686                 if (minPresent == 0) {
 687                     break;
 688                 }
 689             }
 690         }
 691         return (minPresent != 0);
 692     }
 693
 694     /**
 695      * Encapsulates the snapshot bytestring and handles the logic of sending
 696      * snapshot chunks
 697      */
 698     protected class FollowerToSnapshot {
 699         private final ByteString snapshotBytes;
 700         private int offset = 0;
 701         // the next snapshot chunk is sent only if the replyReceivedForOffset matches offset
 702         private int replyReceivedForOffset;
 703         // if replyStatus is false, the previous chunk is attempted
 704         private boolean replyStatus = false;
 705         private int chunkIndex;
 706         private final int totalChunks;
 707         private int lastChunkHashCode = AbstractLeader.INITIAL_LAST_CHUNK_HASH_CODE;
 708         private int nextChunkHashCode = AbstractLeader.INITIAL_LAST_CHUNK_HASH_CODE;
 709
 710         public FollowerToSnapshot(ByteString snapshotBytes) {
 711             this.snapshotBytes = snapshotBytes;
 712             int size = snapshotBytes.size();
 713             totalChunks = ( size / context.getConfigParams().getSnapshotChunkSize()) +
 714                 ((size % context.getConfigParams().getSnapshotChunkSize()) > 0 ? 1 : 0);
 715             if(LOG.isDebugEnabled()) {
 716                 LOG.debug("{}: Snapshot {} bytes, total chunks to send:{}",
 717                         context.getId(), size, totalChunks);
 718             }
 719             replyReceivedForOffset = -1;
 720             chunkIndex = AbstractLeader.FIRST_CHUNK_INDEX;
 721         }
 722
 723         public ByteString getSnapshotBytes() {
 724             return snapshotBytes;
 725         }
 726
 727         public int incrementOffset() {
 728             if(replyStatus) {
 729                 // if prev chunk failed, we would want to sent the same chunk again
 730                 offset = offset + context.getConfigParams().getSnapshotChunkSize();
 731             }
 732             return offset;
 733         }
 734
 735         public int incrementChunkIndex() {
 736             if (replyStatus) {
 737                 // if prev chunk failed, we would want to sent the same chunk again
 738                 chunkIndex =  chunkIndex + 1;
 739             }
 740             return chunkIndex;
 741         }
 742
 743         public int getChunkIndex() {
 744             return chunkIndex;
 745         }
 746
 747         public int getTotalChunks() {
 748             return totalChunks;
 749         }
 750
 751         public boolean canSendNextChunk() {
 752             // we only send a false if a chunk is sent but we have not received a reply yet
 753             return replyReceivedForOffset == offset;
 754         }
 755
 756         public boolean isLastChunk(int chunkIndex) {
 757             return totalChunks == chunkIndex;
 758         }
 759
 760         public void markSendStatus(boolean success) {
 761             if (success) {
 762                 // if the chunk sent was successful
 763                 replyReceivedForOffset = offset;
 764                 replyStatus = true;
 765                 lastChunkHashCode = nextChunkHashCode;
 766             } else {
 767                 // if the chunk sent was failure
 768                 replyReceivedForOffset = offset;
 769                 replyStatus = false;
 770             }
 771         }
 772
 773         public ByteString getNextChunk() {
 774             int snapshotLength = getSnapshotBytes().size();
 775             int start = incrementOffset();
 776             int size = context.getConfigParams().getSnapshotChunkSize();
 777             if (context.getConfigParams().getSnapshotChunkSize() > snapshotLength) {
 778                 size = snapshotLength;
 779             } else {
 780                 if ((start + context.getConfigParams().getSnapshotChunkSize()) > snapshotLength) {
 781                     size = snapshotLength - start;
 782                 }
 783             }
 784
 785             if(LOG.isDebugEnabled()) {
 786                 LOG.debug("{}: Next chunk: length={}, offset={},size={}", context.getId(),
 787                     snapshotLength, start, size);
 788             }
 789             ByteString substring = getSnapshotBytes().substring(start, start + size);
 790             nextChunkHashCode = substring.hashCode();
 791             return substring;
 792         }
 793
 794         /**
 795          * reset should be called when the Follower needs to be sent the snapshot from the beginning
 796          */
 797         public void reset(){
 798             offset = 0;
 799             replyStatus = false;
 800             replyReceivedForOffset = offset;
 801             chunkIndex = AbstractLeader.FIRST_CHUNK_INDEX;
 802             lastChunkHashCode = AbstractLeader.INITIAL_LAST_CHUNK_HASH_CODE;
 803         }
 804
 805         public int getLastChunkHashCode() {
 806             return lastChunkHashCode;
 807         }
 808     }
 809
 810     // called from example-actor for printing the follower-states
 811     public String printFollowerStates() {
 812         final StringBuilder sb = new StringBuilder();
 813
 814         sb.append('[');
 815         for (FollowerLogInformation followerLogInformation : followerToLog.values()) {
 816             sb.append('{');
 817             sb.append(followerLogInformation.getId());
 818             sb.append(" state:");
 819             sb.append(followerLogInformation.isFollowerActive());
 820             sb.append("},");
 821         }
 822         sb.append(']');
 823
 824         return sb.toString();
 825     }
 826
 827     @VisibleForTesting
 828     public FollowerLogInformation getFollower(String followerId) {
 829         return followerToLog.get(followerId);
 830     }
 831
 832     @VisibleForTesting
 833     protected void setFollowerSnapshot(String followerId, FollowerToSnapshot snapshot) {
 834         mapFollowerToSnapshot.put(followerId, snapshot);
 835     }
 836
 837     @VisibleForTesting
 838     public int followerSnapshotSize() {
 839         return mapFollowerToSnapshot.size();
 840     }
 841
 842     @VisibleForTesting
 843     public int followerLogSize() {
 844         return followerToLog.size();
 845     }
 846 }