2 * Copyright (c) 2016 Cisco Systems, Inc. and others. All rights reserved.
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
8 package org.opendaylight.controller.cluster.databroker.actors.dds;
10 import akka.actor.ActorRef;
11 import akka.actor.Status;
12 import com.google.common.base.Throwables;
13 import com.google.common.base.Verify;
14 import java.util.ArrayList;
15 import java.util.Collection;
17 import java.util.concurrent.ConcurrentHashMap;
18 import java.util.concurrent.atomic.AtomicLong;
19 import org.opendaylight.controller.cluster.access.client.BackendInfoResolver;
20 import org.opendaylight.controller.cluster.access.client.ClientActorBehavior;
21 import org.opendaylight.controller.cluster.access.client.ClientActorContext;
22 import org.opendaylight.controller.cluster.access.client.ConnectedClientConnection;
23 import org.opendaylight.controller.cluster.access.concepts.LocalHistoryIdentifier;
24 import org.opendaylight.yangtools.yang.data.api.YangInstanceIdentifier;
25 import org.slf4j.Logger;
26 import org.slf4j.LoggerFactory;
29 * {@link ClientActorBehavior} acting as an intermediary between the backend actors and the DistributedDataStore
33 * This class is not visible outside of this package because it breaks the actor containment. Services provided to
34 * Java world outside of actor containment are captured in {@link DataStoreClient}.
37 * IMPORTANT: this class breaks actor containment via methods implementing {@link DataStoreClient} contract.
38 * When touching internal state, be mindful of the execution context from which execution context, Actor
39 * or POJO, is the state being accessed or modified.
42 * THREAD SAFETY: this class must always be kept thread-safe, so that both the Actor System thread and the application
43 * threads can run concurrently. All state transitions must be made in a thread-safe manner. When in
44 * doubt, feel free to synchronize on this object.
47 * PERFORMANCE: this class lies in a performance-critical fast path. All code needs to be concise and efficient, but
48 * performance must not come at the price of correctness. Any optimizations need to be carefully analyzed
49 * for correctness and performance impact.
52 * TRADE-OFFS: part of the functionality runs in application threads without switching contexts, which makes it ideal
53 * for performing work and charging applications for it. That has two positive effects:
54 * - CPU usage is distributed across applications, minimizing work done in the actor thread
55 * - CPU usage provides back-pressure towards the application.
57 * @author Robert Varga
59 abstract class AbstractDataStoreClientBehavior extends ClientActorBehavior<ShardBackendInfo>
60 implements DataStoreClient {
61 private static final Logger LOG = LoggerFactory.getLogger(AbstractDataStoreClientBehavior.class);
63 private final Map<LocalHistoryIdentifier, ClientLocalHistory> histories = new ConcurrentHashMap<>();
64 private final AtomicLong nextHistoryId = new AtomicLong(1);
65 private final SingleClientHistory singleHistory;
67 private volatile Throwable aborted;
69 AbstractDataStoreClientBehavior(final ClientActorContext context,
70 final BackendInfoResolver<ShardBackendInfo> resolver) {
71 super(context, resolver);
72 singleHistory = new SingleClientHistory(this, new LocalHistoryIdentifier(getIdentifier(), 0));
77 // Methods below are invoked from the client actor thread
82 protected final void haltClient(final Throwable cause) {
83 // If we have encountered a previous problem there is no cleanup necessary, as we have already cleaned up
84 // Thread safely is not an issue, as both this method and any failures are executed from the same (client actor)
86 if (aborted != null) {
87 abortOperations(cause);
91 private void abortOperations(final Throwable cause) {
92 // This acts as a barrier, application threads check this after they have added an entry in the maps,
93 // and if they observe aborted being non-null, they will perform their cleanup and not return the handle.
96 for (ClientLocalHistory h : histories.values()) {
102 private AbstractDataStoreClientBehavior shutdown(final ClientActorBehavior<ShardBackendInfo> currentBehavior) {
103 abortOperations(new IllegalStateException("Client " + getIdentifier() + " has been shut down"));
108 protected final AbstractDataStoreClientBehavior onCommand(final Object command) {
109 if (command instanceof GetClientRequest) {
110 ((GetClientRequest) command).getReplyTo().tell(new Status.Success(this), ActorRef.noSender());
112 LOG.warn("{}: ignoring unhandled command {}", persistenceId(), command);
119 * The connection has resolved, which means we have to potentially perform message adaptation. This is a bit more
120 * involved, as the messages need to be replayed to the individual proxies.
123 protected final ConnectionConnectCohort connectionUp(final ConnectedClientConnection<ShardBackendInfo> newConn) {
124 // Step 1: Freeze all AbstractProxyHistory instances pointing to that shard. This indirectly means that no
125 // further TransactionProxies can be created and we can safely traverse maps without risking
127 final Collection<HistoryReconnectCohort> cohorts = new ArrayList<>();
128 startReconnect(singleHistory, newConn, cohorts);
129 for (ClientLocalHistory h : histories.values()) {
130 startReconnect(h, newConn, cohorts);
133 return previousEntries -> {
135 // Step 2: Collect previous successful requests from the cohorts. We do not want to expose
136 // the non-throttling interface to the connection, hence we use a wrapper consumer
137 for (HistoryReconnectCohort c : cohorts) {
138 c.replaySuccessfulRequests(previousEntries);
141 // Step 3: Install a forwarder, which will forward requests back to affected cohorts. Any outstanding
142 // requests will be immediately sent to it and requests being sent concurrently will get
143 // forwarded once they hit the new connection.
144 return BouncingReconnectForwarder.forCohorts(newConn, cohorts);
146 // Step 4: Complete switchover of the connection. The cohorts can resume normal operations.
147 for (HistoryReconnectCohort c : cohorts) {
154 private static void startReconnect(final AbstractClientHistory history,
155 final ConnectedClientConnection<ShardBackendInfo> newConn,
156 final Collection<HistoryReconnectCohort> cohorts) {
157 final HistoryReconnectCohort cohort = history.startReconnect(newConn);
158 if (cohort != null) {
165 // Methods below are invoked from application threads
170 public final ClientLocalHistory createLocalHistory() {
171 final LocalHistoryIdentifier historyId = new LocalHistoryIdentifier(getIdentifier(),
172 nextHistoryId.getAndIncrement());
173 final ClientLocalHistory history = new ClientLocalHistory(this, historyId);
174 LOG.debug("{}: creating a new local history {}", persistenceId(), history);
176 Verify.verify(histories.put(historyId, history) == null);
178 final Throwable a = aborted;
180 history.localAbort(a);
181 histories.remove(historyId, history);
182 throw Throwables.propagate(a);
189 public final ClientTransaction createTransaction() {
190 return singleHistory.createTransaction();
194 public final void close() {
195 context().executeInActor(this::shutdown);
198 abstract Long resolveShardForPath(final YangInstanceIdentifier path);