opendaylight/md-sal/sal-akka-raft/src/main/java/org/opendaylight/controller/cluster/raft/behaviors/Follower.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.Address;
  14 import akka.cluster.Cluster;
  15 import akka.cluster.ClusterEvent.CurrentClusterState;
  16 import akka.cluster.Member;
  17 import akka.cluster.MemberStatus;
  18 import akka.japi.Procedure;
  19 import com.google.common.annotations.VisibleForTesting;
  20 import com.google.common.base.Stopwatch;
  21 import java.io.IOException;
  22 import java.util.ArrayList;
  23 import java.util.List;
  24 import java.util.Optional;
  25 import java.util.Set;
  26 import java.util.concurrent.TimeUnit;
  27 import java.util.concurrent.atomic.AtomicBoolean;
  28 import javax.annotation.Nullable;
  29 import org.opendaylight.controller.cluster.messaging.MessageAssembler;
  30 import org.opendaylight.controller.cluster.raft.RaftActorContext;
  31 import org.opendaylight.controller.cluster.raft.RaftState;
  32 import org.opendaylight.controller.cluster.raft.ReplicatedLogEntry;
  33 import org.opendaylight.controller.cluster.raft.base.messages.ApplySnapshot;
  34 import org.opendaylight.controller.cluster.raft.base.messages.ElectionTimeout;
  35 import org.opendaylight.controller.cluster.raft.base.messages.TimeoutNow;
  36 import org.opendaylight.controller.cluster.raft.messages.AppendEntries;
  37 import org.opendaylight.controller.cluster.raft.messages.AppendEntriesReply;
  38 import org.opendaylight.controller.cluster.raft.messages.InstallSnapshot;
  39 import org.opendaylight.controller.cluster.raft.messages.InstallSnapshotReply;
  40 import org.opendaylight.controller.cluster.raft.messages.RaftRPC;
  41 import org.opendaylight.controller.cluster.raft.messages.RequestVote;
  42 import org.opendaylight.controller.cluster.raft.messages.RequestVoteReply;
  43 import org.opendaylight.controller.cluster.raft.persisted.ServerConfigurationPayload;
  44 import org.opendaylight.controller.cluster.raft.persisted.Snapshot;
  45
  46 /**
  47  * The behavior of a RaftActor in the Follower raft state.
  48  * <ul>
  49  * <li> Respond to RPCs from candidates and leaders
  50  * <li> If election timeout elapses without receiving AppendEntries
  51  * RPC from current leader or granting vote to candidate:
  52  * convert to candidate
  53  * </ul>
  54  */
  55 public class Follower extends AbstractRaftActorBehavior {
  56     private static final long MAX_ELECTION_TIMEOUT_FACTOR = 18;
  57
  58     private final SyncStatusTracker initialSyncStatusTracker;
  59
  60     private final MessageAssembler appendEntriesMessageAssembler;
  61
  62     private final Stopwatch lastLeaderMessageTimer = Stopwatch.createStarted();
  63     private SnapshotTracker snapshotTracker = null;
  64     private String leaderId;
  65     private short leaderPayloadVersion;
  66
  67     public Follower(final RaftActorContext context) {
  68         this(context, null, (short)-1);
  69     }
  70
  71     public Follower(final RaftActorContext context, final String initialLeaderId,
  72             final short initialLeaderPayloadVersion) {
  73         super(context, RaftState.Follower);
  74         this.leaderId = initialLeaderId;
  75         this.leaderPayloadVersion = initialLeaderPayloadVersion;
  76
  77         initialSyncStatusTracker = new SyncStatusTracker(context.getActor(), getId(), context.getConfigParams()
  78             .getSyncIndexThreshold());
  79
  80         appendEntriesMessageAssembler = MessageAssembler.builder().logContext(logName())
  81                 .fileBackedStreamFactory(context.getFileBackedOutputStreamFactory())
  82                 .assembledMessageCallback((message, sender) -> handleMessage(sender, message)).build();
  83
  84         if (context.getPeerIds().isEmpty() && getLeaderId() == null) {
  85             actor().tell(TimeoutNow.INSTANCE, actor());
  86         } else {
  87             scheduleElection(electionDuration());
  88         }
  89     }
  90
  91     @Override
  92     public final String getLeaderId() {
  93         return leaderId;
  94     }
  95
  96     @VisibleForTesting
  97     protected final void setLeaderId(@Nullable final String leaderId) {
  98         this.leaderId = leaderId;
  99     }
 100
 101     @Override
 102     public short getLeaderPayloadVersion() {
 103         return leaderPayloadVersion;
 104     }
 105
 106     @VisibleForTesting
 107     protected final void setLeaderPayloadVersion(final short leaderPayloadVersion) {
 108         this.leaderPayloadVersion = leaderPayloadVersion;
 109     }
 110
 111     private void restartLastLeaderMessageTimer() {
 112         if (lastLeaderMessageTimer.isRunning()) {
 113             lastLeaderMessageTimer.reset();
 114         }
 115
 116         lastLeaderMessageTimer.start();
 117     }
 118
 119     private boolean isLogEntryPresent(final long index) {
 120         if (context.getReplicatedLog().isInSnapshot(index)) {
 121             return true;
 122         }
 123
 124         ReplicatedLogEntry entry = context.getReplicatedLog().get(index);
 125         return entry != null;
 126
 127     }
 128
 129     private void updateInitialSyncStatus(final long currentLeaderCommit, final String newLeaderId) {
 130         initialSyncStatusTracker.update(newLeaderId, currentLeaderCommit, context.getCommitIndex());
 131     }
 132
 133     @Override
 134     protected RaftActorBehavior handleAppendEntries(final ActorRef sender, final AppendEntries appendEntries) {
 135         int numLogEntries = appendEntries.getEntries().size();
 136         if (log.isTraceEnabled()) {
 137             log.trace("{}: handleAppendEntries: {}", logName(), appendEntries);
 138         } else if (log.isDebugEnabled() && numLogEntries > 0) {
 139             log.debug("{}: handleAppendEntries: {}", logName(), appendEntries);
 140         }
 141
 142         if (snapshotTracker != null && !snapshotTracker.getLeaderId().equals(appendEntries.getLeaderId())) {
 143             log.debug("{}: snapshot install is in progress but the prior snapshot leaderId {} does not match the "
 144                 + "AppendEntries leaderId {}", logName(), snapshotTracker.getLeaderId(), appendEntries.getLeaderId());
 145             closeSnapshotTracker();
 146         }
 147
 148         if (snapshotTracker != null || context.getSnapshotManager().isApplying()) {
 149             // if snapshot install is in progress, follower should just acknowledge append entries with a reply.
 150             AppendEntriesReply reply = new AppendEntriesReply(context.getId(), currentTerm(), true,
 151                     lastIndex(), lastTerm(), context.getPayloadVersion());
 152
 153             log.debug("{}: snapshot install is in progress, replying immediately with {}", logName(), reply);
 154             sender.tell(reply, actor());
 155
 156             return this;
 157         }
 158
 159         // If we got here then we do appear to be talking to the leader
 160         leaderId = appendEntries.getLeaderId();
 161         leaderPayloadVersion = appendEntries.getPayloadVersion();
 162
 163         // First check if the logs are in sync or not
 164         if (isOutOfSync(appendEntries, sender)) {
 165             updateInitialSyncStatus(appendEntries.getLeaderCommit(), appendEntries.getLeaderId());
 166             return this;
 167         }
 168
 169         if (!processNewEntries(appendEntries, sender)) {
 170             updateInitialSyncStatus(appendEntries.getLeaderCommit(), appendEntries.getLeaderId());
 171             return this;
 172         }
 173
 174         long lastIndex = lastIndex();
 175         long prevCommitIndex = context.getCommitIndex();
 176
 177         // If leaderCommit > commitIndex, set commitIndex = min(leaderCommit, index of last new entry)
 178         if (appendEntries.getLeaderCommit() > prevCommitIndex) {
 179             context.setCommitIndex(Math.min(appendEntries.getLeaderCommit(), lastIndex));
 180         }
 181
 182         if (prevCommitIndex != context.getCommitIndex()) {
 183             log.debug("{}: Commit index set to {}", logName(), context.getCommitIndex());
 184         }
 185
 186         AppendEntriesReply reply = new AppendEntriesReply(context.getId(), currentTerm(), true,
 187                 lastIndex, lastTerm(), context.getPayloadVersion());
 188
 189         if (log.isTraceEnabled()) {
 190             log.trace("{}: handleAppendEntries returning : {}", logName(), reply);
 191         } else if (log.isDebugEnabled() && numLogEntries > 0) {
 192             log.debug("{}: handleAppendEntries returning : {}", logName(), reply);
 193         }
 194
 195         // Reply to the leader before applying any previous state so as not to hold up leader consensus.
 196         sender.tell(reply, actor());
 197
 198         updateInitialSyncStatus(appendEntries.getLeaderCommit(), appendEntries.getLeaderId());
 199
 200         // If leaderCommit > lastApplied, increment lastApplied and apply log[lastApplied] to state machine (§5.3).
 201         // lastApplied can be equal to lastIndex.
 202         if (appendEntries.getLeaderCommit() > context.getLastApplied() && context.getLastApplied() < lastIndex) {
 203             if (log.isDebugEnabled()) {
 204                 log.debug("{}: applyLogToStateMachine, appendEntries.getLeaderCommit(): {}, "
 205                         + "context.getLastApplied(): {}, lastIndex(): {}", logName(),
 206                     appendEntries.getLeaderCommit(), context.getLastApplied(), lastIndex);
 207             }
 208
 209             applyLogToStateMachine(appendEntries.getLeaderCommit());
 210         }
 211
 212         if (!context.getSnapshotManager().isCapturing()) {
 213             super.performSnapshotWithoutCapture(appendEntries.getReplicatedToAllIndex());
 214         }
 215
 216         appendEntriesMessageAssembler.checkExpiredAssembledMessageState();
 217
 218         return this;
 219     }
 220
 221     private boolean processNewEntries(final AppendEntries appendEntries, final ActorRef sender) {
 222         int numLogEntries = appendEntries.getEntries().size();
 223         if (numLogEntries == 0) {
 224             return true;
 225         }
 226
 227         log.debug("{}: Number of entries to be appended = {}", logName(), numLogEntries);
 228
 229         long lastIndex = lastIndex();
 230         int addEntriesFrom = 0;
 231
 232         // First check for conflicting entries. If an existing entry conflicts with a new one (same index but different
 233         // term), delete the existing entry and all that follow it (§5.3)
 234         if (context.getReplicatedLog().size() > 0) {
 235             // Find the entry up until the one that is not in the follower's log
 236             for (int i = 0;i < numLogEntries; i++, addEntriesFrom++) {
 237                 ReplicatedLogEntry matchEntry = appendEntries.getEntries().get(i);
 238
 239                 if (!isLogEntryPresent(matchEntry.getIndex())) {
 240                     // newEntry not found in the log
 241                     break;
 242                 }
 243
 244                 long existingEntryTerm = getLogEntryTerm(matchEntry.getIndex());
 245
 246                 log.debug("{}: matchEntry {} is present: existingEntryTerm: {}", logName(), matchEntry,
 247                         existingEntryTerm);
 248
 249                 // existingEntryTerm == -1 means it's in the snapshot and not in the log. We don't know
 250                 // what the term was so we'll assume it matches.
 251                 if (existingEntryTerm == -1 || existingEntryTerm == matchEntry.getTerm()) {
 252                     continue;
 253                 }
 254
 255                 if (!context.getRaftPolicy().applyModificationToStateBeforeConsensus()) {
 256                     log.info("{}: Removing entries from log starting at {}, commitIndex: {}, lastApplied: {}",
 257                             logName(), matchEntry.getIndex(), context.getCommitIndex(), context.getLastApplied());
 258
 259                     // Entries do not match so remove all subsequent entries but only if the existing entries haven't
 260                     // been applied to the state yet.
 261                     if (matchEntry.getIndex() <= context.getLastApplied()
 262                             || !context.getReplicatedLog().removeFromAndPersist(matchEntry.getIndex())) {
 263                         // Could not remove the entries - this means the matchEntry index must be in the
 264                         // snapshot and not the log. In this case the prior entries are part of the state
 265                         // so we must send back a reply to force a snapshot to completely re-sync the
 266                         // follower's log and state.
 267
 268                         log.info("{}: Could not remove entries - sending reply to force snapshot", logName());
 269                         sender.tell(new AppendEntriesReply(context.getId(), currentTerm(), false, lastIndex,
 270                                 lastTerm(), context.getPayloadVersion(), true), actor());
 271                         return false;
 272                     }
 273
 274                     break;
 275                 } else {
 276                     sender.tell(new AppendEntriesReply(context.getId(), currentTerm(), false, lastIndex,
 277                             lastTerm(), context.getPayloadVersion(), true), actor());
 278                     return false;
 279                 }
 280             }
 281         }
 282
 283         lastIndex = lastIndex();
 284         log.debug("{}: After cleanup, lastIndex: {}, entries to be added from: {}", logName(), lastIndex,
 285                 addEntriesFrom);
 286
 287         // When persistence successfully completes for each new log entry appended, we need to determine if we
 288         // should capture a snapshot to compact the persisted log. shouldCaptureSnapshot tracks whether or not
 289         // one of the log entries has exceeded the log size threshold whereby a snapshot should be taken. However
 290         // we don't initiate the snapshot at that log entry but rather after the last log entry has been persisted.
 291         // This is done because subsequent log entries after the one that tripped the threshold may have been
 292         // applied to the state already, as the persistence callback occurs async, and we want those entries
 293         // purged from the persisted log as well.
 294         final AtomicBoolean shouldCaptureSnapshot = new AtomicBoolean(false);
 295         final Procedure<ReplicatedLogEntry> appendAndPersistCallback = logEntry -> {
 296             final List<ReplicatedLogEntry> entries = appendEntries.getEntries();
 297             final ReplicatedLogEntry lastEntryToAppend = entries.get(entries.size() - 1);
 298             if (shouldCaptureSnapshot.get() && logEntry == lastEntryToAppend) {
 299                 context.getSnapshotManager().capture(context.getReplicatedLog().last(), getReplicatedToAllIndex());
 300             }
 301         };
 302
 303         // Append any new entries not already in the log
 304         for (int i = addEntriesFrom; i < numLogEntries; i++) {
 305             ReplicatedLogEntry entry = appendEntries.getEntries().get(i);
 306
 307             log.debug("{}: Append entry to log {}", logName(), entry.getData());
 308
 309             context.getReplicatedLog().appendAndPersist(entry, appendAndPersistCallback, false);
 310
 311             shouldCaptureSnapshot.compareAndSet(false,
 312                     context.getReplicatedLog().shouldCaptureSnapshot(entry.getIndex()));
 313
 314             if (entry.getData() instanceof ServerConfigurationPayload) {
 315                 context.updatePeerIds((ServerConfigurationPayload)entry.getData());
 316             }
 317         }
 318
 319         log.debug("{}: Log size is now {}", logName(), context.getReplicatedLog().size());
 320
 321         return true;
 322     }
 323
 324     private boolean isOutOfSync(final AppendEntries appendEntries, final ActorRef sender) {
 325
 326         final long lastIndex = lastIndex();
 327         if (lastIndex == -1 && appendEntries.getPrevLogIndex() != -1) {
 328
 329             // The follower's log is out of sync because the leader does have an entry at prevLogIndex and this
 330             // follower has no entries in it's log.
 331
 332             log.info("{}: The followers log is empty and the senders prevLogIndex is {}", logName(),
 333                 appendEntries.getPrevLogIndex());
 334
 335             sendOutOfSyncAppendEntriesReply(sender, false);
 336             return true;
 337         }
 338
 339         if (lastIndex > -1) {
 340             if (isLogEntryPresent(appendEntries.getPrevLogIndex())) {
 341                 final long leadersPrevLogTermInFollowersLogOrSnapshot =
 342                         getLogEntryOrSnapshotTerm(appendEntries.getPrevLogIndex());
 343                 if (leadersPrevLogTermInFollowersLogOrSnapshot != appendEntries.getPrevLogTerm()) {
 344
 345                     // The follower's log is out of sync because the Leader's prevLogIndex entry does exist
 346                     // in the follower's log or snapshot but it has a different term.
 347
 348                     log.info("{}: The prevLogIndex {} was found in the log but the term {} is not equal to the append "
 349                         + "entries prevLogTerm {} - lastIndex: {}, snapshotIndex: {}, snapshotTerm: {}", logName(),
 350                         appendEntries.getPrevLogIndex(), leadersPrevLogTermInFollowersLogOrSnapshot,
 351                         appendEntries.getPrevLogTerm(), lastIndex, context.getReplicatedLog().getSnapshotIndex(),
 352                         context.getReplicatedLog().getSnapshotTerm());
 353
 354                     sendOutOfSyncAppendEntriesReply(sender, false);
 355                     return true;
 356                 }
 357             } else if (appendEntries.getPrevLogIndex() != -1) {
 358
 359                 // The follower's log is out of sync because the Leader's prevLogIndex entry was not found in it's log
 360
 361                 log.info("{}: The log is not empty but the prevLogIndex {} was not found in it - lastIndex: {}, "
 362                         + "snapshotIndex: {}, snapshotTerm: {}", logName(), appendEntries.getPrevLogIndex(), lastIndex,
 363                         context.getReplicatedLog().getSnapshotIndex(), context.getReplicatedLog().getSnapshotTerm());
 364
 365                 sendOutOfSyncAppendEntriesReply(sender, false);
 366                 return true;
 367             }
 368         }
 369
 370         if (appendEntries.getPrevLogIndex() == -1 && appendEntries.getPrevLogTerm() == -1
 371                 && appendEntries.getReplicatedToAllIndex() != -1) {
 372             if (!isLogEntryPresent(appendEntries.getReplicatedToAllIndex())) {
 373                 // This append entry comes from a leader who has it's log aggressively trimmed and so does not have
 374                 // the previous entry in it's in-memory journal
 375
 376                 log.info("{}: Cannot append entries because the replicatedToAllIndex {} does not appear to be in the "
 377                         + "in-memory journal - lastIndex: {}, snapshotIndex: {}, snapshotTerm: {}", logName(),
 378                         appendEntries.getReplicatedToAllIndex(), lastIndex,
 379                         context.getReplicatedLog().getSnapshotIndex(), context.getReplicatedLog().getSnapshotTerm());
 380
 381                 sendOutOfSyncAppendEntriesReply(sender, false);
 382                 return true;
 383             }
 384
 385             final List<ReplicatedLogEntry> entries = appendEntries.getEntries();
 386             if (entries.size() > 0 && !isLogEntryPresent(entries.get(0).getIndex() - 1)) {
 387                 log.info("{}: Cannot append entries because the calculated previousIndex {} was not found in the "
 388                         + "in-memory journal - lastIndex: {}, snapshotIndex: {}, snapshotTerm: {}", logName(),
 389                         entries.get(0).getIndex() - 1, lastIndex, context.getReplicatedLog().getSnapshotIndex(),
 390                         context.getReplicatedLog().getSnapshotTerm());
 391
 392                 sendOutOfSyncAppendEntriesReply(sender, false);
 393                 return true;
 394             }
 395         }
 396
 397         return false;
 398     }
 399
 400     private void sendOutOfSyncAppendEntriesReply(final ActorRef sender, boolean forceInstallSnapshot) {
 401         // We found that the log was out of sync so just send a negative reply.
 402         final AppendEntriesReply reply = new AppendEntriesReply(context.getId(), currentTerm(), false, lastIndex(),
 403                 lastTerm(), context.getPayloadVersion(), forceInstallSnapshot);
 404
 405         log.info("{}: Follower is out-of-sync so sending negative reply: {}", logName(), reply);
 406         sender.tell(reply, actor());
 407     }
 408
 409     @Override
 410     protected RaftActorBehavior handleAppendEntriesReply(final ActorRef sender,
 411         final AppendEntriesReply appendEntriesReply) {
 412         return this;
 413     }
 414
 415     @Override
 416     protected RaftActorBehavior handleRequestVoteReply(final ActorRef sender,
 417         final RequestVoteReply requestVoteReply) {
 418         return this;
 419     }
 420
 421     @Override
 422     public RaftActorBehavior handleMessage(final ActorRef sender, final Object message) {
 423         if (message instanceof ElectionTimeout || message instanceof TimeoutNow) {
 424             return handleElectionTimeout(message);
 425         }
 426
 427         if (appendEntriesMessageAssembler.handleMessage(message, actor())) {
 428             return this;
 429         }
 430
 431         if (!(message instanceof RaftRPC)) {
 432             // The rest of the processing requires the message to be a RaftRPC
 433             return null;
 434         }
 435
 436         final RaftRPC rpc = (RaftRPC) message;
 437         // If RPC request or response contains term T > currentTerm:
 438         // set currentTerm = T, convert to follower (§5.1)
 439         // This applies to all RPC messages and responses
 440         if (rpc.getTerm() > context.getTermInformation().getCurrentTerm()) {
 441             log.info("{}: Term {} in \"{}\" message is greater than follower's term {} - updating term",
 442                 logName(), rpc.getTerm(), rpc, context.getTermInformation().getCurrentTerm());
 443
 444             context.getTermInformation().updateAndPersist(rpc.getTerm(), null);
 445         }
 446
 447         if (rpc instanceof InstallSnapshot) {
 448             handleInstallSnapshot(sender, (InstallSnapshot) rpc);
 449             restartLastLeaderMessageTimer();
 450             scheduleElection(electionDuration());
 451             return this;
 452         }
 453
 454         if (!(rpc instanceof RequestVote) || canGrantVote((RequestVote) rpc)) {
 455             restartLastLeaderMessageTimer();
 456             scheduleElection(electionDuration());
 457         }
 458
 459         return super.handleMessage(sender, rpc);
 460     }
 461
 462     private RaftActorBehavior handleElectionTimeout(final Object message) {
 463         // If the message is ElectionTimeout, verify we haven't actually seen a message from the leader
 464         // during the election timeout interval. It may that the election timer expired b/c this actor
 465         // was busy and messages got delayed, in which case leader messages would be backed up in the
 466         // queue but would be processed before the ElectionTimeout message and thus would restart the
 467         // lastLeaderMessageTimer.
 468         long lastLeaderMessageInterval = lastLeaderMessageTimer.elapsed(TimeUnit.MILLISECONDS);
 469         long electionTimeoutInMillis = context.getConfigParams().getElectionTimeOutInterval().toMillis();
 470         boolean noLeaderMessageReceived = !lastLeaderMessageTimer.isRunning()
 471                 || lastLeaderMessageInterval >= electionTimeoutInMillis;
 472
 473         if (canStartElection()) {
 474             if (message instanceof TimeoutNow) {
 475                 log.debug("{}: Received TimeoutNow - switching to Candidate", logName());
 476                 return internalSwitchBehavior(RaftState.Candidate);
 477             } else if (noLeaderMessageReceived) {
 478                 // Check the cluster state to see if the leader is known to be up before we go to Candidate.
 479                 // However if we haven't heard from the leader in a long time even though the cluster state
 480                 // indicates it's up then something is wrong - leader might be stuck indefinitely - so switch
 481                 // to Candidate,
 482                 long maxElectionTimeout = electionTimeoutInMillis * MAX_ELECTION_TIMEOUT_FACTOR;
 483                 if (isLeaderAvailabilityKnown() && lastLeaderMessageInterval < maxElectionTimeout) {
 484                     log.debug("{}: Received ElectionTimeout but leader appears to be available", logName());
 485                     scheduleElection(electionDuration());
 486                 } else {
 487                     log.debug("{}: Received ElectionTimeout - switching to Candidate", logName());
 488                     return internalSwitchBehavior(RaftState.Candidate);
 489                 }
 490             } else {
 491                 log.debug("{}: Received ElectionTimeout but lastLeaderMessageInterval {} < election timeout {}",
 492                         logName(), lastLeaderMessageInterval, context.getConfigParams().getElectionTimeOutInterval());
 493                 scheduleElection(electionDuration());
 494             }
 495         } else if (message instanceof ElectionTimeout) {
 496             if (noLeaderMessageReceived) {
 497                 setLeaderId(null);
 498             }
 499
 500             scheduleElection(electionDuration());
 501         }
 502
 503         return this;
 504     }
 505
 506     private boolean isLeaderAvailabilityKnown() {
 507         if (leaderId == null) {
 508             return false;
 509         }
 510
 511         Optional<Cluster> cluster = context.getCluster();
 512         if (!cluster.isPresent()) {
 513             return false;
 514         }
 515
 516         ActorSelection leaderActor = context.getPeerActorSelection(leaderId);
 517         if (leaderActor == null) {
 518             return false;
 519         }
 520
 521         Address leaderAddress = leaderActor.anchorPath().address();
 522
 523         CurrentClusterState state = cluster.get().state();
 524         Set<Member> unreachable = state.getUnreachable();
 525
 526         log.debug("{}: Checking for leader {} in the cluster unreachable set {}", logName(), leaderAddress,
 527                 unreachable);
 528
 529         for (Member m: unreachable) {
 530             if (leaderAddress.equals(m.address())) {
 531                 log.info("{}: Leader {} is unreachable", logName(), leaderAddress);
 532                 return false;
 533             }
 534         }
 535
 536         for (Member m: state.getMembers()) {
 537             if (leaderAddress.equals(m.address())) {
 538                 if (m.status() == MemberStatus.up() || m.status() == MemberStatus.weaklyUp()) {
 539                     log.debug("{}: Leader {} cluster status is {} - leader is available", logName(),
 540                             leaderAddress, m.status());
 541                     return true;
 542                 } else {
 543                     log.debug("{}: Leader {} cluster status is {} - leader is unavailable", logName(),
 544                             leaderAddress, m.status());
 545                     return false;
 546                 }
 547             }
 548         }
 549
 550         log.debug("{}: Leader {} not found in the cluster member set", logName(), leaderAddress);
 551
 552         return false;
 553     }
 554
 555     private void handleInstallSnapshot(final ActorRef sender, final InstallSnapshot installSnapshot) {
 556
 557         log.debug("{}: handleInstallSnapshot: {}", logName(), installSnapshot);
 558
 559         leaderId = installSnapshot.getLeaderId();
 560
 561         if (snapshotTracker == null) {
 562             snapshotTracker = new SnapshotTracker(log, installSnapshot.getTotalChunks(), installSnapshot.getLeaderId(),
 563                     context);
 564         }
 565
 566         updateInitialSyncStatus(installSnapshot.getLastIncludedIndex(), installSnapshot.getLeaderId());
 567
 568         try {
 569             final InstallSnapshotReply reply = new InstallSnapshotReply(
 570                     currentTerm(), context.getId(), installSnapshot.getChunkIndex(), true);
 571
 572             if (snapshotTracker.addChunk(installSnapshot.getChunkIndex(), installSnapshot.getData(),
 573                     installSnapshot.getLastChunkHashCode())) {
 574
 575                 log.info("{}: Snapshot installed from leader: {}", logName(), installSnapshot.getLeaderId());
 576
 577                 Snapshot snapshot = Snapshot.create(
 578                         context.getSnapshotManager().convertSnapshot(snapshotTracker.getSnapshotBytes()),
 579                         new ArrayList<>(),
 580                         installSnapshot.getLastIncludedIndex(),
 581                         installSnapshot.getLastIncludedTerm(),
 582                         installSnapshot.getLastIncludedIndex(),
 583                         installSnapshot.getLastIncludedTerm(),
 584                         context.getTermInformation().getCurrentTerm(),
 585                         context.getTermInformation().getVotedFor(),
 586                         installSnapshot.getServerConfig().orNull());
 587
 588                 ApplySnapshot.Callback applySnapshotCallback = new ApplySnapshot.Callback() {
 589                     @Override
 590                     public void onSuccess() {
 591                         log.debug("{}: handleInstallSnapshot returning: {}", logName(), reply);
 592
 593                         sender.tell(reply, actor());
 594                     }
 595
 596                     @Override
 597                     public void onFailure() {
 598                         sender.tell(new InstallSnapshotReply(currentTerm(), context.getId(), -1, false), actor());
 599                     }
 600                 };
 601
 602                 actor().tell(new ApplySnapshot(snapshot, applySnapshotCallback), actor());
 603
 604                 closeSnapshotTracker();
 605             } else {
 606                 log.debug("{}: handleInstallSnapshot returning: {}", logName(), reply);
 607
 608                 sender.tell(reply, actor());
 609             }
 610         } catch (IOException e) {
 611             log.debug("{}: Exception in InstallSnapshot of follower", logName(), e);
 612
 613             sender.tell(new InstallSnapshotReply(currentTerm(), context.getId(),
 614                     -1, false), actor());
 615
 616             closeSnapshotTracker();
 617         }
 618     }
 619
 620     private void closeSnapshotTracker() {
 621         if (snapshotTracker != null) {
 622             snapshotTracker.close();
 623             snapshotTracker = null;
 624         }
 625     }
 626
 627     @Override
 628     public void close() {
 629         closeSnapshotTracker();
 630         stopElection();
 631         appendEntriesMessageAssembler.close();
 632     }
 633
 634     @VisibleForTesting
 635     SnapshotTracker getSnapshotTracker() {
 636         return snapshotTracker;
 637     }
 638 }