opendaylight/md-sal/cds-access-client/src/main/java/org/opendaylight/controller/cluster/access/client/ClientActorBehavior.java

   1 /*
   2  * Copyright (c) 2016 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 package org.opendaylight.controller.cluster.access.client;
   9
  10 import com.google.common.annotations.Beta;
  11 import com.google.common.base.Preconditions;
  12 import com.google.common.base.Stopwatch;
  13 import com.google.common.base.Verify;
  14 import java.util.Collection;
  15 import java.util.Map;
  16 import java.util.Optional;
  17 import java.util.concurrent.ConcurrentHashMap;
  18 import java.util.concurrent.TimeUnit;
  19 import java.util.concurrent.TimeoutException;
  20 import javax.annotation.Nonnull;
  21 import javax.annotation.Nullable;
  22 import javax.annotation.concurrent.GuardedBy;
  23 import org.opendaylight.controller.cluster.access.commands.NotLeaderException;
  24 import org.opendaylight.controller.cluster.access.commands.OutOfSequenceEnvelopeException;
  25 import org.opendaylight.controller.cluster.access.concepts.ClientIdentifier;
  26 import org.opendaylight.controller.cluster.access.concepts.FailureEnvelope;
  27 import org.opendaylight.controller.cluster.access.concepts.LocalHistoryIdentifier;
  28 import org.opendaylight.controller.cluster.access.concepts.RequestException;
  29 import org.opendaylight.controller.cluster.access.concepts.RequestFailure;
  30 import org.opendaylight.controller.cluster.access.concepts.ResponseEnvelope;
  31 import org.opendaylight.controller.cluster.access.concepts.RetiredGenerationException;
  32 import org.opendaylight.controller.cluster.access.concepts.RuntimeRequestException;
  33 import org.opendaylight.controller.cluster.access.concepts.SuccessEnvelope;
  34 import org.opendaylight.controller.cluster.access.concepts.TransactionIdentifier;
  35 import org.opendaylight.controller.cluster.common.actor.Dispatchers.DispatcherType;
  36 import org.opendaylight.controller.cluster.io.FileBackedOutputStreamFactory;
  37 import org.opendaylight.controller.cluster.messaging.MessageAssembler;
  38 import org.opendaylight.yangtools.concepts.Identifiable;
  39 import org.opendaylight.yangtools.concepts.WritableIdentifier;
  40 import org.slf4j.Logger;
  41 import org.slf4j.LoggerFactory;
  42 import scala.concurrent.duration.FiniteDuration;
  43
  44 /**
  45  * A behavior, which handles messages sent to a {@link AbstractClientActor}.
  46  *
  47  * @author Robert Varga
  48  */
  49 @Beta
  50 public abstract class ClientActorBehavior<T extends BackendInfo> extends
  51         RecoveredClientActorBehavior<ClientActorContext> implements Identifiable<ClientIdentifier> {
  52     /**
  53      * Connection reconnect cohort, driven by this class.
  54      */
  55     @FunctionalInterface
  56     protected interface ConnectionConnectCohort {
  57         /**
  58          * Finish the connection by replaying previous messages onto the new connection.
  59          *
  60          * @param enqueuedEntries Previously-enqueued entries
  61          * @return A {@link ReconnectForwarder} to handle any straggler messages which arrive after this method returns.
  62          */
  63         @Nonnull ReconnectForwarder finishReconnect(@Nonnull Collection<ConnectionEntry> enqueuedEntries);
  64     }
  65
  66     private static final Logger LOG = LoggerFactory.getLogger(ClientActorBehavior.class);
  67     private static final FiniteDuration RESOLVE_RETRY_DURATION = FiniteDuration.apply(5, TimeUnit.SECONDS);
  68
  69     /**
  70      * Map of connections to the backend. This map is concurrent to allow lookups, but given complex operations
  71      * involved in connection transitions it is protected by a {@link InversibleLock}. Write-side of the lock is taken
  72      * during connection transitions. Optimistic read-side of the lock is taken when new connections are introduced
  73      * into the map.
  74      *
  75      * <p>
  76      * The lock detects potential AB/BA deadlock scenarios and will force the reader side out by throwing
  77      * a {@link InversibleLockException} -- which must be propagated up, releasing locks as it propagates. The initial
  78      * entry point causing the the conflicting lookup must then call {@link InversibleLockException#awaitResolution()}
  79      * before retrying the operation.
  80      */
  81     // TODO: it should be possible to move these two into ClientActorContext
  82     private final Map<Long, AbstractClientConnection<T>> connections = new ConcurrentHashMap<>();
  83     private final InversibleLock connectionsLock = new InversibleLock();
  84     private final BackendInfoResolver<T> resolver;
  85     private final MessageAssembler responseMessageAssembler;
  86
  87     protected ClientActorBehavior(@Nonnull final ClientActorContext context,
  88             @Nonnull final BackendInfoResolver<T> resolver) {
  89         super(context);
  90         this.resolver = Preconditions.checkNotNull(resolver);
  91
  92         final ClientActorConfig config = context.config();
  93         responseMessageAssembler = MessageAssembler.builder().logContext(persistenceId())
  94                 .fileBackedStreamFactory(new FileBackedOutputStreamFactory(config.getFileBackedStreamingThreshold(),
  95                         config.getTempFileDirectory()))
  96                 .assembledMessageCallback((message, sender) -> context.self().tell(message, sender)).build();
  97     }
  98
  99     @Override
 100     @Nonnull
 101     public final ClientIdentifier getIdentifier() {
 102         return context().getIdentifier();
 103     }
 104
 105     @Override
 106     public void close() {
 107         responseMessageAssembler.close();
 108     }
 109
 110     /**
 111      * Get a connection to a shard.
 112      *
 113      * @param shard Shard cookie
 114      * @return Connection to a shard
 115      * @throws InversibleLockException if the shard is being reconnected
 116      */
 117     public final AbstractClientConnection<T> getConnection(final Long shard) {
 118         while (true) {
 119             final long stamp = connectionsLock.optimisticRead();
 120             final AbstractClientConnection<T> conn = connections.computeIfAbsent(shard, this::createConnection);
 121             if (connectionsLock.validate(stamp)) {
 122                 // No write-lock in-between, return success
 123                 return conn;
 124             }
 125         }
 126     }
 127
 128     private AbstractClientConnection<T> getConnection(final ResponseEnvelope<?> response) {
 129         // Always called from actor context: no locking required
 130         return connections.get(extractCookie(response.getMessage().getTarget()));
 131     }
 132
 133     @SuppressWarnings("unchecked")
 134     @Override
 135     final ClientActorBehavior<T> onReceiveCommand(final Object command) {
 136         if (command instanceof InternalCommand) {
 137             return ((InternalCommand<T>) command).execute(this);
 138         }
 139
 140         if (command instanceof SuccessEnvelope) {
 141             return onRequestSuccess((SuccessEnvelope) command);
 142         }
 143
 144         if (command instanceof FailureEnvelope) {
 145             return internalOnRequestFailure((FailureEnvelope) command);
 146         }
 147
 148         if (MessageAssembler.isHandledMessage(command)) {
 149             context().dispatchers().getDispatcher(DispatcherType.Serialization).execute(
 150                 () -> responseMessageAssembler.handleMessage(command, context().self()));
 151             return this;
 152         }
 153
 154         return onCommand(command);
 155     }
 156
 157     private static long extractCookie(final WritableIdentifier id) {
 158         if (id instanceof TransactionIdentifier) {
 159             return ((TransactionIdentifier) id).getHistoryId().getCookie();
 160         } else if (id instanceof LocalHistoryIdentifier) {
 161             return ((LocalHistoryIdentifier) id).getCookie();
 162         } else {
 163             throw new IllegalArgumentException("Unhandled identifier " + id);
 164         }
 165     }
 166
 167     private void onResponse(final ResponseEnvelope<?> response) {
 168         final AbstractClientConnection<T> connection = getConnection(response);
 169         if (connection != null) {
 170             connection.receiveResponse(response);
 171         } else {
 172             LOG.info("{}: Ignoring unknown response {}", persistenceId(), response);
 173         }
 174     }
 175
 176     private ClientActorBehavior<T> onRequestSuccess(final SuccessEnvelope success) {
 177         onResponse(success);
 178         return this;
 179     }
 180
 181     private ClientActorBehavior<T> onRequestFailure(final FailureEnvelope failure) {
 182         onResponse(failure);
 183         return this;
 184     }
 185
 186     private ClientActorBehavior<T> internalOnRequestFailure(final FailureEnvelope command) {
 187         final AbstractClientConnection<T> conn = getConnection(command);
 188         if (conn != null) {
 189             /*
 190              * We are talking to multiple actors, which may be lagging behind our state significantly. This has
 191              * the effect that we may be receiving responses from a previous connection after we have created a new
 192              * one to a different actor.
 193              *
 194              * Since we are already replaying requests to the new actor, we want to ignore errors reported on the old
 195              * connection -- for example NotLeaderException, which must not cause a new reconnect. Check the envelope's
 196              * sessionId and if it does not match our current connection just ignore it.
 197              */
 198             final Optional<T> optBackend = conn.getBackendInfo();
 199             if (optBackend.isPresent() && optBackend.get().getSessionId() != command.getSessionId()) {
 200                 LOG.debug("{}: Mismatched current connection {} and envelope {}, ignoring response", persistenceId(),
 201                     conn, command);
 202                 return this;
 203             }
 204         }
 205
 206         final RequestFailure<?, ?> failure = command.getMessage();
 207         final RequestException cause = failure.getCause();
 208         if (cause instanceof RetiredGenerationException) {
 209             LOG.error("{}: current generation {} has been superseded", persistenceId(), getIdentifier(), cause);
 210             haltClient(cause);
 211             poison(cause);
 212             return null;
 213         }
 214         if (cause instanceof NotLeaderException) {
 215             if (conn instanceof ReconnectingClientConnection) {
 216                 // Already reconnecting, do not churn the logs
 217                 return this;
 218             } else if (conn != null) {
 219                 LOG.info("{}: connection {} indicated no leadership, reconnecting it", persistenceId(), conn, cause);
 220                 return conn.reconnect(this, cause);
 221             }
 222         }
 223         if (cause instanceof OutOfSequenceEnvelopeException) {
 224             if (conn instanceof ReconnectingClientConnection) {
 225                 // Already reconnecting, do not churn the logs
 226                 return this;
 227             } else if (conn != null) {
 228                 LOG.info("{}: connection {} indicated sequencing mismatch on {} sequence {} ({}), reconnecting it",
 229                     persistenceId(), conn, failure.getTarget(), failure.getSequence(), command.getTxSequence(), cause);
 230                 return conn.reconnect(this, cause);
 231             }
 232         }
 233
 234         return onRequestFailure(command);
 235     }
 236
 237     private void poison(final RequestException cause) {
 238         final long stamp = connectionsLock.writeLock();
 239         try {
 240             for (AbstractClientConnection<T> q : connections.values()) {
 241                 q.poison(cause);
 242             }
 243
 244             connections.clear();
 245         } finally {
 246             connectionsLock.unlockWrite(stamp);
 247         }
 248     }
 249
 250     /**
 251      * Halt And Catch Fire. Halt processing on this client. Implementations need to ensure they initiate state flush
 252      * procedures. No attempt to use this instance should be made after this method returns. Any such use may result
 253      * in undefined behavior.
 254      *
 255      * @param cause Failure cause
 256      */
 257     protected abstract void haltClient(@Nonnull Throwable cause);
 258
 259     /**
 260      * Override this method to handle any command which is not handled by the base behavior.
 261      *
 262      * @param command the command to process
 263      * @return Next behavior to use, null if this actor should shut down.
 264      */
 265     @Nullable
 266     protected abstract ClientActorBehavior<T> onCommand(@Nonnull Object command);
 267
 268     /**
 269      * Override this method to provide a backend resolver instance.
 270      *
 271      * @return a backend resolver instance
 272      */
 273     protected final @Nonnull BackendInfoResolver<T> resolver() {
 274         return resolver;
 275     }
 276
 277     /**
 278      * Callback invoked when a new connection has been established. Implementations are expected perform preparatory
 279      * tasks before the previous connection is frozen.
 280      *
 281      * @param newConn New connection
 282      * @return ConnectionConnectCohort which will be used to complete the process of bringing the connection up.
 283      */
 284     @GuardedBy("connectionsLock")
 285     @Nonnull protected abstract ConnectionConnectCohort connectionUp(@Nonnull ConnectedClientConnection<T> newConn);
 286
 287     private void backendConnectFinished(final Long shard, final AbstractClientConnection<T> conn,
 288             final T backend, final Throwable failure) {
 289         if (failure != null) {
 290             if (failure instanceof TimeoutException) {
 291                 if (!conn.equals(connections.get(shard))) {
 292                     // AbstractClientConnection will remove itself when it decides there is no point in continuing,
 293                     // at which point we want to stop retrying
 294                     LOG.info("{}: stopping resolution of shard {} on stale connection {}", persistenceId(), shard, conn,
 295                         failure);
 296                     return;
 297                 }
 298
 299                 LOG.debug("{}: timed out resolving shard {}, scheduling retry in {}", persistenceId(), shard,
 300                     RESOLVE_RETRY_DURATION, failure);
 301                 context().executeInActor(b -> {
 302                     resolveConnection(shard, conn);
 303                     return b;
 304                 }, RESOLVE_RETRY_DURATION);
 305                 return;
 306             }
 307
 308             LOG.error("{}: failed to resolve shard {}", persistenceId(), shard, failure);
 309             final RequestException cause;
 310             if (failure instanceof RequestException) {
 311                 cause = (RequestException) failure;
 312             } else {
 313                 cause = new RuntimeRequestException("Failed to resolve shard " + shard, failure);
 314             }
 315
 316             conn.poison(cause);
 317             return;
 318         }
 319
 320         LOG.info("{}: resolved shard {} to {}", persistenceId(), shard, backend);
 321         final long stamp = connectionsLock.writeLock();
 322         try {
 323             final Stopwatch sw = Stopwatch.createStarted();
 324
 325             // Create a new connected connection
 326             final ConnectedClientConnection<T> newConn = new ConnectedClientConnection<>(conn.context(),
 327                     conn.cookie(), backend);
 328             LOG.info("{}: resolving connection {} to {}", persistenceId(), conn, newConn);
 329
 330             // Start reconnecting without the old connection lock held
 331             final ConnectionConnectCohort cohort = Verify.verifyNotNull(connectionUp(newConn));
 332
 333             // Lock the old connection and get a reference to its entries
 334             final Collection<ConnectionEntry> replayIterable = conn.startReplay();
 335
 336             // Finish the connection attempt
 337             final ReconnectForwarder forwarder = Verify.verifyNotNull(cohort.finishReconnect(replayIterable));
 338
 339             // Install the forwarder, unlocking the old connection
 340             conn.finishReplay(forwarder);
 341
 342             // Make sure new lookups pick up the new connection
 343             if (!connections.replace(shard, conn, newConn)) {
 344                 final AbstractClientConnection<T> existing = connections.get(conn.cookie());
 345                 LOG.warn("{}: old connection {} does not match existing {}, new connection {} in limbo",
 346                     persistenceId(), conn, existing, newConn);
 347             } else {
 348                 LOG.info("{}: replaced connection {} with {} in {}", persistenceId(), conn, newConn, sw);
 349             }
 350         } finally {
 351             connectionsLock.unlockWrite(stamp);
 352         }
 353     }
 354
 355     void removeConnection(final AbstractClientConnection<?> conn) {
 356         final long stamp = connectionsLock.writeLock();
 357         try {
 358             if (!connections.remove(conn.cookie(), conn)) {
 359                 final AbstractClientConnection<T> existing = connections.get(conn.cookie());
 360                 if (existing != null) {
 361                     LOG.warn("{}: failed to remove connection {}, as it was superseded by {}", persistenceId(), conn,
 362                         existing);
 363                 } else {
 364                     LOG.warn("{}: failed to remove connection {}, as it was not tracked", persistenceId(), conn);
 365                 }
 366             } else {
 367                 LOG.info("{}: removed connection {}", persistenceId(), conn);
 368             }
 369         } finally {
 370             connectionsLock.unlockWrite(stamp);
 371         }
 372     }
 373
 374     @SuppressWarnings("unchecked")
 375     void reconnectConnection(final ConnectedClientConnection<?> oldConn,
 376             final ReconnectingClientConnection<?> newConn) {
 377         final ReconnectingClientConnection<T> conn = (ReconnectingClientConnection<T>)newConn;
 378         LOG.info("{}: connection {} reconnecting as {}", persistenceId(), oldConn, newConn);
 379
 380         final long stamp = connectionsLock.writeLock();
 381         try {
 382             final boolean replaced = connections.replace(oldConn.cookie(), (AbstractClientConnection<T>)oldConn, conn);
 383             if (!replaced) {
 384                 final AbstractClientConnection<T> existing = connections.get(oldConn.cookie());
 385                 if (existing != null) {
 386                     LOG.warn("{}: failed to replace connection {}, as it was superseded by {}", persistenceId(), conn,
 387                         existing);
 388                 } else {
 389                     LOG.warn("{}: failed to replace connection {}, as it was not tracked", persistenceId(), conn);
 390                 }
 391             }
 392         } finally {
 393             connectionsLock.unlockWrite(stamp);
 394         }
 395
 396         final Long shard = oldConn.cookie();
 397         LOG.info("{}: refreshing backend for shard {}", persistenceId(), shard);
 398         resolver().refreshBackendInfo(shard, conn.getBackendInfo().get()).whenComplete(
 399             (backend, failure) -> context().executeInActor(behavior -> {
 400                 backendConnectFinished(shard, conn, backend, failure);
 401                 return behavior;
 402             }));
 403     }
 404
 405     private ConnectingClientConnection<T> createConnection(final Long shard) {
 406         final ConnectingClientConnection<T> conn = new ConnectingClientConnection<>(context(), shard);
 407         resolveConnection(shard, conn);
 408         return conn;
 409     }
 410
 411     private void resolveConnection(final Long shard, final AbstractClientConnection<T> conn) {
 412         LOG.debug("{}: resolving shard {} connection {}", persistenceId(), shard, conn);
 413         resolver().getBackendInfo(shard).whenComplete((backend, failure) -> context().executeInActor(behavior -> {
 414             backendConnectFinished(shard, conn, backend, failure);
 415             return behavior;
 416         }));
 417     }
 418 }