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.japi.Procedure;
  13 import com.google.common.annotations.VisibleForTesting;
  14 import com.google.common.base.Preconditions;
  15 import java.util.ArrayList;
  16 import org.opendaylight.controller.cluster.raft.RaftActorContext;
  17 import org.opendaylight.controller.cluster.raft.RaftState;
  18 import org.opendaylight.controller.cluster.raft.ReplicatedLogEntry;
  19 import org.opendaylight.controller.cluster.raft.ServerConfigurationPayload;
  20 import org.opendaylight.controller.cluster.raft.Snapshot;
  21 import org.opendaylight.controller.cluster.raft.base.messages.ApplySnapshot;
  22 import org.opendaylight.controller.cluster.raft.base.messages.ElectionTimeout;
  23 import org.opendaylight.controller.cluster.raft.messages.AppendEntries;
  24 import org.opendaylight.controller.cluster.raft.messages.AppendEntriesReply;
  25 import org.opendaylight.controller.cluster.raft.messages.InstallSnapshot;
  26 import org.opendaylight.controller.cluster.raft.messages.InstallSnapshotReply;
  27 import org.opendaylight.controller.cluster.raft.messages.RaftRPC;
  28 import org.opendaylight.controller.cluster.raft.messages.RequestVote;
  29 import org.opendaylight.controller.cluster.raft.messages.RequestVoteReply;
  30
  31 /**
  32  * The behavior of a RaftActor in the Follower state
  33  * <p/>
  34  * <ul>
  35  * <li> Respond to RPCs from candidates and leaders
  36  * <li> If election timeout elapses without receiving AppendEntries
  37  * RPC from current leader or granting vote to candidate:
  38  * convert to candidate
  39  * </ul>
  40  */
  41 public class Follower extends AbstractRaftActorBehavior {
  42     private static final int SYNC_THRESHOLD = 10;
  43
  44     private final SyncStatusTracker initialSyncStatusTracker;
  45
  46     private final Procedure<ReplicatedLogEntry> appendAndPersistCallback = new Procedure<ReplicatedLogEntry>() {
  47         @Override
  48         public void apply(ReplicatedLogEntry logEntry) {
  49             context.getReplicatedLog().captureSnapshotIfReady(logEntry);
  50         }
  51     };
  52
  53     private SnapshotTracker snapshotTracker = null;
  54     private String leaderId;
  55     private short leaderPayloadVersion;
  56
  57     public Follower(RaftActorContext context) {
  58         this(context, null, (short)-1);
  59     }
  60
  61     public Follower(RaftActorContext context, String initialLeaderId, short initialLeaderPayloadVersion) {
  62         super(context, RaftState.Follower);
  63         this.leaderId = initialLeaderId;
  64         this.leaderPayloadVersion = initialLeaderPayloadVersion;
  65
  66         initialSyncStatusTracker = new SyncStatusTracker(context.getActor(), getId(), SYNC_THRESHOLD);
  67
  68         if(canStartElection()) {
  69             if (context.getPeerIds().isEmpty() && getLeaderId() == null) {
  70                 actor().tell(ElectionTimeout.INSTANCE, actor());
  71             } else {
  72                 scheduleElection(electionDuration());
  73             }
  74         }
  75     }
  76
  77     @Override
  78     public final String getLeaderId() {
  79         return leaderId;
  80     }
  81
  82     @VisibleForTesting
  83     protected final void setLeaderId(final String leaderId) {
  84         this.leaderId = Preconditions.checkNotNull(leaderId);
  85     }
  86
  87     @Override
  88     public short getLeaderPayloadVersion() {
  89         return leaderPayloadVersion;
  90     }
  91
  92     @VisibleForTesting
  93     protected final void setLeaderPayloadVersion(short leaderPayloadVersion) {
  94         this.leaderPayloadVersion = leaderPayloadVersion;
  95     }
  96
  97     private boolean isLogEntryPresent(long index){
  98         if(index == context.getReplicatedLog().getSnapshotIndex()){
  99             return true;
 100         }
 101
 102         ReplicatedLogEntry previousEntry = context.getReplicatedLog()
 103                 .get(index);
 104
 105         return previousEntry != null;
 106
 107     }
 108
 109     private long getLogEntryTerm(long index){
 110         if(index == context.getReplicatedLog().getSnapshotIndex()){
 111             return context.getReplicatedLog().getSnapshotTerm();
 112         }
 113
 114         ReplicatedLogEntry previousEntry = context.getReplicatedLog()
 115                 .get(index);
 116
 117         if(previousEntry != null){
 118             return previousEntry.getTerm();
 119         }
 120
 121         return -1;
 122     }
 123
 124     private void updateInitialSyncStatus(long currentLeaderCommit, String leaderId){
 125         initialSyncStatusTracker.update(leaderId, currentLeaderCommit, context.getCommitIndex());
 126     }
 127
 128     @Override
 129     protected RaftActorBehavior handleAppendEntries(ActorRef sender, AppendEntries appendEntries) {
 130
 131         int numLogEntries = appendEntries.getEntries() != null ? appendEntries.getEntries().size() : 0;
 132         if(LOG.isTraceEnabled()) {
 133             LOG.trace("{}: handleAppendEntries: {}", logName(), appendEntries);
 134         } else if(LOG.isDebugEnabled() && numLogEntries > 0) {
 135             LOG.debug("{}: handleAppendEntries: {}", logName(), appendEntries);
 136         }
 137
 138         // TODO : Refactor this method into a bunch of smaller methods
 139         // to make it easier to read. Before refactoring ensure tests
 140         // cover the code properly
 141
 142         if (snapshotTracker != null || context.getSnapshotManager().isApplying()) {
 143             // if snapshot install is in progress, follower should just acknowledge append entries with a reply.
 144             AppendEntriesReply reply = new AppendEntriesReply(context.getId(), currentTerm(), true,
 145                     lastIndex(), lastTerm(), context.getPayloadVersion());
 146
 147             if(LOG.isDebugEnabled()) {
 148                 LOG.debug("{}: snapshot install is in progress, replying immediately with {}", logName(), reply);
 149             }
 150             sender.tell(reply, actor());
 151
 152             return this;
 153         }
 154
 155         // If we got here then we do appear to be talking to the leader
 156         leaderId = appendEntries.getLeaderId();
 157         leaderPayloadVersion = appendEntries.getPayloadVersion();
 158
 159         updateInitialSyncStatus(appendEntries.getLeaderCommit(), appendEntries.getLeaderId());
 160         // First check if the logs are in sync or not
 161         long lastIndex = lastIndex();
 162
 163         if (isOutOfSync(appendEntries)) {
 164             // We found that the log was out of sync so just send a negative
 165             // reply and return
 166
 167             LOG.debug("{}: Follower is out-of-sync, so sending negative reply, lastIndex: {}, lastTerm: {}",
 168                         logName(), lastIndex, lastTerm());
 169
 170             sender.tell(new AppendEntriesReply(context.getId(), currentTerm(), false, lastIndex,
 171                     lastTerm(), context.getPayloadVersion()), actor());
 172             return this;
 173         }
 174
 175         if (appendEntries.getEntries() != null && appendEntries.getEntries().size() > 0) {
 176
 177             LOG.debug("{}: Number of entries to be appended = {}", logName(),
 178                         appendEntries.getEntries().size());
 179
 180             // 3. If an existing entry conflicts with a new one (same index
 181             // but different terms), delete the existing entry and all that
 182             // follow it (§5.3)
 183             int addEntriesFrom = 0;
 184             if (context.getReplicatedLog().size() > 0) {
 185
 186                 // Find the entry up until the one that is not in the follower's log
 187                 for (int i = 0;i < appendEntries.getEntries().size(); i++, addEntriesFrom++) {
 188                     ReplicatedLogEntry matchEntry = appendEntries.getEntries().get(i);
 189                     ReplicatedLogEntry newEntry = context.getReplicatedLog().get(matchEntry.getIndex());
 190
 191                     if (newEntry == null) {
 192                         //newEntry not found in the log
 193                         break;
 194                     }
 195
 196                     if (newEntry.getTerm() == matchEntry.getTerm()) {
 197                         continue;
 198                     }
 199
 200                     if(!context.getRaftPolicy().applyModificationToStateBeforeConsensus()) {
 201
 202                         LOG.debug("{}: Removing entries from log starting at {}", logName(),
 203                                 matchEntry.getIndex());
 204
 205                         // Entries do not match so remove all subsequent entries
 206                         context.getReplicatedLog().removeFromAndPersist(matchEntry.getIndex());
 207                         break;
 208                     } else {
 209                         sender.tell(new AppendEntriesReply(context.getId(), currentTerm(), false, lastIndex,
 210                                 lastTerm(), context.getPayloadVersion(), true), actor());
 211                         return this;
 212                     }
 213                 }
 214             }
 215
 216             lastIndex = lastIndex();
 217             LOG.debug("{}: After cleanup entries to be added from = {}", logName(),
 218                         (addEntriesFrom + lastIndex));
 219
 220             // 4. Append any new entries not already in the log
 221             for (int i = addEntriesFrom; i < appendEntries.getEntries().size(); i++) {
 222                 ReplicatedLogEntry entry = appendEntries.getEntries().get(i);
 223
 224                 LOG.debug("{}: Append entry to log {}", logName(), entry.getData());
 225
 226                 context.getReplicatedLog().appendAndPersist(entry, appendAndPersistCallback);
 227
 228                 if(entry.getData() instanceof ServerConfigurationPayload) {
 229                     context.updatePeerIds((ServerConfigurationPayload)entry.getData());
 230                 }
 231             }
 232
 233             LOG.debug("{}: Log size is now {}", logName(), context.getReplicatedLog().size());
 234         }
 235
 236         // 5. If leaderCommit > commitIndex, set commitIndex =
 237         // min(leaderCommit, index of last new entry)
 238
 239         lastIndex = lastIndex();
 240         long prevCommitIndex = context.getCommitIndex();
 241
 242         context.setCommitIndex(Math.min(appendEntries.getLeaderCommit(), lastIndex));
 243
 244         if (prevCommitIndex != context.getCommitIndex()) {
 245             LOG.debug("{}: Commit index set to {}", logName(), context.getCommitIndex());
 246         }
 247
 248         // If commitIndex > lastApplied: increment lastApplied, apply
 249         // log[lastApplied] to state machine (§5.3)
 250         // check if there are any entries to be applied. last-applied can be equal to last-index
 251         if (appendEntries.getLeaderCommit() > context.getLastApplied() &&
 252             context.getLastApplied() < lastIndex) {
 253             if(LOG.isDebugEnabled()) {
 254                 LOG.debug("{}: applyLogToStateMachine, " +
 255                         "appendEntries.getLeaderCommit(): {}," +
 256                         "context.getLastApplied(): {}, lastIndex(): {}", logName(),
 257                     appendEntries.getLeaderCommit(), context.getLastApplied(), lastIndex);
 258             }
 259
 260             applyLogToStateMachine(appendEntries.getLeaderCommit());
 261         }
 262
 263         AppendEntriesReply reply = new AppendEntriesReply(context.getId(), currentTerm(), true,
 264             lastIndex, lastTerm(), context.getPayloadVersion());
 265
 266         if(LOG.isTraceEnabled()) {
 267             LOG.trace("{}: handleAppendEntries returning : {}", logName(), reply);
 268         } else if(LOG.isDebugEnabled() && numLogEntries > 0) {
 269             LOG.debug("{}: handleAppendEntries returning : {}", logName(), reply);
 270         }
 271
 272         sender.tell(reply, actor());
 273
 274         if (!context.getSnapshotManager().isCapturing()) {
 275             super.performSnapshotWithoutCapture(appendEntries.getReplicatedToAllIndex());
 276         }
 277
 278         return this;
 279     }
 280
 281     private boolean isOutOfSync(AppendEntries appendEntries) {
 282
 283         long prevLogTerm = getLogEntryTerm(appendEntries.getPrevLogIndex());
 284         boolean prevEntryPresent = isLogEntryPresent(appendEntries.getPrevLogIndex());
 285         long lastIndex = lastIndex();
 286         int numLogEntries = appendEntries.getEntries() != null ? appendEntries.getEntries().size() : 0;
 287         boolean outOfSync = true;
 288
 289         if (lastIndex == -1 && appendEntries.getPrevLogIndex() != -1) {
 290
 291             // The follower's log is out of sync because the leader does have
 292             // an entry at prevLogIndex and this follower has no entries in
 293             // it's log.
 294
 295             LOG.debug("{}: The followers log is empty and the senders prevLogIndex is {}",
 296                         logName(), appendEntries.getPrevLogIndex());
 297         } else if (lastIndex > -1 && appendEntries.getPrevLogIndex() != -1 && !prevEntryPresent) {
 298
 299             // The follower's log is out of sync because the Leader's
 300             // prevLogIndex entry was not found in it's log
 301
 302             LOG.debug("{}: The log is not empty but the prevLogIndex {} was not found in it",
 303                         logName(), appendEntries.getPrevLogIndex());
 304         } else if (lastIndex > -1 && prevEntryPresent && prevLogTerm != appendEntries.getPrevLogTerm()) {
 305
 306             // The follower's log is out of sync because the Leader's
 307             // prevLogIndex entry does exist in the follower's log but it has
 308             // a different term in it
 309
 310             LOG.debug(
 311                     "{}: Cannot append entries because previous entry term {}  is not equal to append entries prevLogTerm {}",
 312                     logName(), prevLogTerm, appendEntries.getPrevLogTerm());
 313         } else if(appendEntries.getPrevLogIndex() == -1 && appendEntries.getPrevLogTerm() == -1
 314                 && appendEntries.getReplicatedToAllIndex() != -1
 315                 && !isLogEntryPresent(appendEntries.getReplicatedToAllIndex())
 316                 && !context.getReplicatedLog().isInSnapshot(appendEntries.getReplicatedToAllIndex())) {
 317             // This append entry comes from a leader who has it's log aggressively trimmed and so does not have
 318             // the previous entry in it's in-memory journal
 319
 320             LOG.debug(
 321                     "{}: Cannot append entries because the replicatedToAllIndex {} does not appear to be in the in-memory journal",
 322                     logName(), appendEntries.getReplicatedToAllIndex());
 323         } else if(appendEntries.getPrevLogIndex() == -1 && appendEntries.getPrevLogTerm() == -1
 324                 && appendEntries.getReplicatedToAllIndex() != -1 && numLogEntries > 0
 325                 && !isLogEntryPresent(appendEntries.getEntries().get(0).getIndex() - 1)
 326                 && !context.getReplicatedLog().isInSnapshot(appendEntries.getEntries().get(0).getIndex() - 1)) {
 327             LOG.debug(
 328                     "{}: Cannot append entries because the calculated previousIndex {} was not found in the in-memory journal",
 329                     logName(), appendEntries.getEntries().get(0).getIndex() - 1);
 330         } else {
 331             outOfSync = false;
 332         }
 333         return outOfSync;
 334     }
 335
 336     @Override
 337     protected RaftActorBehavior handleAppendEntriesReply(ActorRef sender,
 338         AppendEntriesReply appendEntriesReply) {
 339         return this;
 340     }
 341
 342     @Override
 343     protected RaftActorBehavior handleRequestVoteReply(ActorRef sender,
 344         RequestVoteReply requestVoteReply) {
 345         return this;
 346     }
 347
 348     @Override
 349     public RaftActorBehavior handleMessage(ActorRef sender, Object originalMessage) {
 350         if (originalMessage instanceof ElectionTimeout) {
 351             if (canStartElection()) {
 352                 LOG.debug("{}: Received ElectionTimeout - switching to Candidate", logName());
 353                 return internalSwitchBehavior(RaftState.Candidate);
 354             } else {
 355                 return this;
 356             }
 357         }
 358
 359         final Object message = fromSerializableMessage(originalMessage);
 360         if (message instanceof RaftRPC) {
 361             RaftRPC rpc = (RaftRPC) message;
 362             // If RPC request or response contains term T > currentTerm:
 363             // set currentTerm = T, convert to follower (§5.1)
 364             // This applies to all RPC messages and responses
 365             if (rpc.getTerm() > context.getTermInformation().getCurrentTerm()) {
 366                 LOG.debug("{}: Term {} in \"{}\" message is greater than follower's term {} - updating term",
 367                         logName(), rpc.getTerm(), rpc, context.getTermInformation().getCurrentTerm());
 368
 369                 context.getTermInformation().updateAndPersist(rpc.getTerm(), null);
 370             }
 371         }
 372
 373         if (message instanceof InstallSnapshot) {
 374             InstallSnapshot installSnapshot = (InstallSnapshot) message;
 375             handleInstallSnapshot(sender, installSnapshot);
 376         }
 377
 378         if (message instanceof RaftRPC && (!(message instanceof RequestVote) || (canGrantVote((RequestVote) message)))){
 379             scheduleElection(electionDuration());
 380         }
 381
 382         return super.handleMessage(sender, message);
 383     }
 384
 385     private void handleInstallSnapshot(final ActorRef sender, InstallSnapshot installSnapshot) {
 386
 387         LOG.debug("{}: handleInstallSnapshot: {}", logName(), installSnapshot);
 388
 389         leaderId = installSnapshot.getLeaderId();
 390
 391         if(snapshotTracker == null){
 392             snapshotTracker = new SnapshotTracker(LOG, installSnapshot.getTotalChunks());
 393         }
 394
 395         updateInitialSyncStatus(installSnapshot.getLastIncludedIndex(), installSnapshot.getLeaderId());
 396
 397         try {
 398             final InstallSnapshotReply reply = new InstallSnapshotReply(
 399                     currentTerm(), context.getId(), installSnapshot.getChunkIndex(), true);
 400
 401             if(snapshotTracker.addChunk(installSnapshot.getChunkIndex(), installSnapshot.getData(),
 402                     installSnapshot.getLastChunkHashCode())){
 403                 Snapshot snapshot = Snapshot.create(snapshotTracker.getSnapshot(),
 404                         new ArrayList<ReplicatedLogEntry>(),
 405                         installSnapshot.getLastIncludedIndex(),
 406                         installSnapshot.getLastIncludedTerm(),
 407                         installSnapshot.getLastIncludedIndex(),
 408                         installSnapshot.getLastIncludedTerm(),
 409                         context.getTermInformation().getCurrentTerm(),
 410                         context.getTermInformation().getVotedFor(),
 411                         context.getPeerServerInfo(true));
 412
 413                 ApplySnapshot.Callback applySnapshotCallback = new ApplySnapshot.Callback() {
 414                     @Override
 415                     public void onSuccess() {
 416                         LOG.debug("{}: handleInstallSnapshot returning: {}", logName(), reply);
 417
 418                         sender.tell(reply, actor());
 419                     }
 420
 421                     @Override
 422                     public void onFailure() {
 423                         sender.tell(new InstallSnapshotReply(currentTerm(), context.getId(), -1, false), actor());
 424                     }
 425                 };
 426
 427                 actor().tell(new ApplySnapshot(snapshot, applySnapshotCallback), actor());
 428
 429                 snapshotTracker = null;
 430             } else {
 431                 LOG.debug("{}: handleInstallSnapshot returning: {}", logName(), reply);
 432
 433                 sender.tell(reply, actor());
 434             }
 435         } catch (SnapshotTracker.InvalidChunkException e) {
 436             LOG.debug("{}: Exception in InstallSnapshot of follower", logName(), e);
 437
 438             sender.tell(new InstallSnapshotReply(currentTerm(), context.getId(),
 439                     -1, false), actor());
 440             snapshotTracker = null;
 441
 442         } catch (Exception e){
 443             LOG.error("{}: Exception in InstallSnapshot of follower", logName(), e);
 444
 445             //send reply with success as false. The chunk will be sent again on failure
 446             sender.tell(new InstallSnapshotReply(currentTerm(), context.getId(),
 447                     installSnapshot.getChunkIndex(), false), actor());
 448
 449         }
 450     }
 451
 452     @Override
 453     public void close() {
 454         stopElection();
 455     }
 456
 457     @VisibleForTesting
 458     SnapshotTracker getSnapshotTracker(){
 459         return snapshotTracker;
 460     }
 461 }