2 * Copyright (c) 2014 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.md.sal.dom.broker.impl;
10 import com.google.common.base.Optional;
11 import com.google.common.base.Preconditions;
12 import com.google.common.base.Verify;
13 import com.google.common.util.concurrent.CheckedFuture;
14 import com.google.common.util.concurrent.FutureCallback;
15 import com.google.common.util.concurrent.Futures;
16 import com.google.common.util.concurrent.ListenableFuture;
17 import com.google.common.util.concurrent.MoreExecutors;
18 import java.util.AbstractMap.SimpleImmutableEntry;
19 import java.util.Map.Entry;
20 import java.util.concurrent.CancellationException;
21 import java.util.concurrent.atomic.AtomicReferenceFieldUpdater;
22 import javax.annotation.Nonnull;
23 import javax.annotation.concurrent.GuardedBy;
24 import org.opendaylight.controller.md.sal.common.api.TransactionStatus;
25 import org.opendaylight.controller.md.sal.common.api.data.AsyncTransaction;
26 import org.opendaylight.controller.md.sal.common.api.data.LogicalDatastoreType;
27 import org.opendaylight.controller.md.sal.common.api.data.ReadFailedException;
28 import org.opendaylight.controller.md.sal.common.api.data.TransactionChain;
29 import org.opendaylight.controller.md.sal.common.api.data.TransactionChainListener;
30 import org.opendaylight.controller.md.sal.common.api.data.TransactionCommitFailedException;
31 import org.opendaylight.controller.md.sal.dom.api.DOMDataBroker;
32 import org.opendaylight.controller.md.sal.dom.api.DOMDataReadOnlyTransaction;
33 import org.opendaylight.controller.md.sal.dom.api.DOMDataReadWriteTransaction;
34 import org.opendaylight.controller.md.sal.dom.api.DOMDataWriteTransaction;
35 import org.opendaylight.controller.md.sal.dom.api.DOMTransactionChain;
36 import org.opendaylight.controller.md.sal.dom.spi.ForwardingDOMDataReadWriteTransaction;
37 import org.opendaylight.yangtools.yang.common.RpcResult;
38 import org.opendaylight.yangtools.yang.data.api.YangInstanceIdentifier;
39 import org.opendaylight.yangtools.yang.data.api.schema.NormalizedNode;
40 import org.slf4j.Logger;
41 import org.slf4j.LoggerFactory;
44 * An implementation of {@link DOMTransactionChain}, which has a very specific
45 * behavior, which some users may find surprising. If keeps the general
46 * intent of the contract, but it makes sure there are never more than two
47 * transactions allocated at any given time: one of them is being committed,
48 * and while that is happening, the other one acts as the scratch pad. Once
49 * the committing transaction completes successfully, the scratch transaction
50 * is enqueued as soon as it is ready.
52 * This mode of operation means that there is no inherent isolation between
53 * the front-end transactions and transactions cannot be reasonably cancelled.
55 * It furthermore means that the transactions returned by {@link #newReadOnlyTransaction()}
56 * counts as an outstanding transaction and the user may not allocate multiple
57 * read-only transactions at the same time.
59 public final class PingPongTransactionChain implements DOMTransactionChain {
60 private static final Logger LOG = LoggerFactory.getLogger(PingPongTransactionChain.class);
61 private final TransactionChainListener listener;
62 private final DOMTransactionChain delegate;
65 private boolean failed;
67 private PingPongTransaction shutdownTx;
69 private Entry<PingPongTransaction, Throwable> deadTx;
72 * This updater is used to manipulate the "ready" transaction. We perform only atomic
75 private static final AtomicReferenceFieldUpdater<PingPongTransactionChain, PingPongTransaction> READY_UPDATER =
76 AtomicReferenceFieldUpdater.newUpdater(PingPongTransactionChain.class, PingPongTransaction.class, "readyTx");
77 private volatile PingPongTransaction readyTx;
80 * This updater is used to manipulate the "locked" transaction. A locked transaction
81 * means we know that the user still holds a transaction and should at some point call
82 * us. We perform on compare-and-swap to ensure we properly detect when a user is
83 * attempting to allocated multiple transactions concurrently.
85 private static final AtomicReferenceFieldUpdater<PingPongTransactionChain, PingPongTransaction> LOCKED_UPDATER =
86 AtomicReferenceFieldUpdater.newUpdater(PingPongTransactionChain.class, PingPongTransaction.class, "lockedTx");
87 private volatile PingPongTransaction lockedTx;
90 * This updater is used to manipulate the "inflight" transaction. There can be at most
91 * one of these at any given time. We perform only compare-and-swap on these.
93 private static final AtomicReferenceFieldUpdater<PingPongTransactionChain, PingPongTransaction> INFLIGHT_UPDATER =
94 AtomicReferenceFieldUpdater.newUpdater(PingPongTransactionChain.class, PingPongTransaction.class, "inflightTx");
95 private volatile PingPongTransaction inflightTx;
97 PingPongTransactionChain(final DOMDataBroker broker, final TransactionChainListener listener) {
98 this.listener = Preconditions.checkNotNull(listener);
99 this.delegate = broker.createTransactionChain(new TransactionChainListener() {
101 public void onTransactionChainFailed(final TransactionChain<?, ?> chain,
102 final AsyncTransaction<?, ?> transaction, final Throwable cause) {
103 LOG.debug("Transaction chain {} reported failure in {}", chain, transaction, cause);
104 delegateFailed(chain, cause);
108 public void onTransactionChainSuccessful(final TransactionChain<?, ?> chain) {
109 delegateSuccessful(chain);
114 void delegateSuccessful(final TransactionChain<?, ?> chain) {
115 final Entry<PingPongTransaction, Throwable> canceled;
116 synchronized (this) {
117 // This looks weird, but we need not hold the lock while invoking callbacks
121 if (canceled == null) {
122 listener.onTransactionChainSuccessful(this);
126 // Backend shutdown successful, but we have a batch of transactions we have to report as dead due to the
127 // user calling cancel().
128 final PingPongTransaction tx = canceled.getKey();
129 final Throwable cause = canceled.getValue();
130 LOG.debug("Transaction chain {} successful, failing cancelled transaction {}", chain, tx, cause);
132 listener.onTransactionChainFailed(this, tx.getFrontendTransaction(), cause);
136 void delegateFailed(final TransactionChain<?, ?> chain, final Throwable cause) {
138 final DOMDataReadWriteTransaction frontend;
139 final PingPongTransaction tx = inflightTx;
141 LOG.warn("Transaction chain {} failed with no pending transactions", chain);
144 frontend = tx.getFrontendTransaction();
147 listener.onTransactionChainFailed(this, frontend, cause);
149 synchronized (this) {
153 * If we do not have a locked transaction, we need to ensure that
154 * the backend transaction is cancelled. Otherwise we can defer
155 * until the user calls us.
157 if (lockedTx == null) {
163 private synchronized PingPongTransaction slowAllocateTransaction() {
164 Preconditions.checkState(shutdownTx == null, "Transaction chain %s has been shut down", this);
166 if (deadTx != null) {
167 throw new IllegalStateException(String.format(
168 "Transaction chain %s has failed due to transaction %s being canceled", this, deadTx.getKey()),
172 final DOMDataReadWriteTransaction delegateTx = delegate.newReadWriteTransaction();
173 final PingPongTransaction newTx = new PingPongTransaction(delegateTx);
175 if (!LOCKED_UPDATER.compareAndSet(this, null, newTx)) {
177 throw new IllegalStateException(String.format("New transaction %s raced with transaction %s", newTx, lockedTx));
183 private PingPongTransaction allocateTransaction() {
184 // Step 1: acquire current state
185 final PingPongTransaction oldTx = READY_UPDATER.getAndSet(this, null);
187 // Slow path: allocate a delegate transaction
189 return slowAllocateTransaction();
192 // Fast path: reuse current transaction. We will check failures and similar on submit().
193 if (!LOCKED_UPDATER.compareAndSet(this, null, oldTx)) {
194 // Ouch. Delegate chain has not detected a duplicate transaction allocation. This is the best we can do.
195 oldTx.getTransaction().cancel();
196 throw new IllegalStateException(String.format("Reusable transaction %s raced with transaction %s", oldTx, lockedTx));
203 * This forces allocateTransaction() on a slow path, which has to happen after
204 * this method has completed executing. Also inflightTx may be updated outside
205 * the lock, hence we need to re-check.
208 private void processIfReady() {
209 if (inflightTx == null) {
210 final PingPongTransaction tx = READY_UPDATER.getAndSet(this, null);
212 processTransaction(tx);
218 * Process a ready transaction. The caller needs to ensure that
219 * each transaction is seen only once by this method.
221 * @param tx Transaction which needs processing.
224 private void processTransaction(@Nonnull final PingPongTransaction tx) {
226 LOG.debug("Cancelling transaction {}", tx);
227 tx.getTransaction().cancel();
231 LOG.debug("Submitting transaction {}", tx);
232 if (!INFLIGHT_UPDATER.compareAndSet(this, null, tx)) {
233 LOG.warn("Submitting transaction {} while {} is still running", tx, inflightTx);
236 Futures.addCallback(tx.getTransaction().submit(), new FutureCallback<Void>() {
238 public void onSuccess(final Void result) {
239 transactionSuccessful(tx, result);
243 public void onFailure(final Throwable t) {
244 transactionFailed(tx, t);
246 }, MoreExecutors.directExecutor());
250 * We got invoked from the data store thread. We need to do two things:
251 * 1) release the in-flight transaction
252 * 2) process the potential next transaction
254 * We have to perform 2) under lock. We could perform 1) without locking, but that means the CAS result may
255 * not be accurate, as a user thread may submit the ready transaction before we acquire the lock -- and checking
256 * for next transaction is not enough, as that may have also be allocated (as a result of a quick
257 * submit/allocate/submit between 1) and 2)). Hence we'd end up doing the following:
258 * 1) CAS of inflightTx
260 * 3) volatile read of inflightTx
262 * Rather than doing that, we keep this method synchronized, hence performing only:
264 * 2) CAS of inflightTx
266 * Since the user thread is barred from submitting the transaction (in processIfReady), we can then proceed with
267 * the knowledge that inflightTx is null -- processTransaction() will still do a CAS, but that is only for
270 private synchronized void processNextTransaction(final PingPongTransaction tx) {
271 final boolean success = INFLIGHT_UPDATER.compareAndSet(this, tx, null);
272 Preconditions.checkState(success, "Completed transaction %s while %s was submitted", tx, inflightTx);
274 final PingPongTransaction nextTx = READY_UPDATER.getAndSet(this, null);
275 if (nextTx != null) {
276 processTransaction(nextTx);
277 } else if (shutdownTx != null) {
278 processTransaction(shutdownTx);
284 void transactionSuccessful(final PingPongTransaction tx, final Void result) {
285 LOG.debug("Transaction {} completed successfully", tx);
287 tx.onSuccess(result);
288 processNextTransaction(tx);
291 void transactionFailed(final PingPongTransaction tx, final Throwable t) {
292 LOG.debug("Transaction {} failed", tx, t);
295 processNextTransaction(tx);
298 void readyTransaction(@Nonnull final PingPongTransaction tx) {
299 // First mark the transaction as not locked.
300 final boolean lockedMatch = LOCKED_UPDATER.compareAndSet(this, tx, null);
301 Preconditions.checkState(lockedMatch, "Attempted to submit transaction %s while we have %s", tx, lockedTx);
302 LOG.debug("Transaction {} unlocked", tx);
305 * The transaction is ready. It will then be picked up by either next allocation,
306 * or a background transaction completion callback.
308 final boolean success = READY_UPDATER.compareAndSet(this, null, tx);
309 Preconditions.checkState(success, "Transaction %s collided on ready state", tx, readyTx);
310 LOG.debug("Transaction {} readied", tx);
313 * We do not see a transaction being in-flight, so we need to take care of dispatching
314 * the transaction to the backend. We are in the ready case, we cannot short-cut
315 * the checking of readyTx, as an in-flight transaction may have completed between us
316 * setting the field above and us checking.
318 if (inflightTx == null) {
319 synchronized (this) {
326 * Transaction cancellation is a heavyweight operation. We only support cancelation of a locked transaction
327 * and return false for everything else. Cancelling such a transaction will result in all transactions in the
328 * batch to be cancelled.
330 * @param tx Backend shared transaction
332 * @param isOpen indicator whether the transaction was already closed
334 synchronized void cancelTransaction(final PingPongTransaction tx, final DOMDataReadWriteTransaction frontendTx) {
335 // Attempt to unlock the operation.
336 final boolean lockedMatch = LOCKED_UPDATER.compareAndSet(this, tx, null);
337 Verify.verify(lockedMatch, "Cancelling transaction %s collided with locked transaction %s", tx, lockedTx);
339 // Cancel the backend transaction, so we do not end up leaking it.
340 final boolean backendCancelled = tx.getTransaction().cancel();
343 // The transaction has failed, this is probably the user just clearing up the transaction they had. We have
344 // already cancelled the transaction anyway,
346 } else if (!backendCancelled) {
347 LOG.warn("Backend transaction cannot be cancelled during cancellation of {}, attempting to continue", tx);
350 // We have dealt with canceling the backend transaction and have unlocked the transaction. Since we are still
351 // inside the synchronized block, any allocations are blocking on the slow path. Now we have to decide the fate
352 // of this transaction chain.
354 // If there are no other frontend transactions in this batch we are aligned with backend state and we can
355 // continue processing.
356 if (frontendTx.equals(tx.getFrontendTransaction())) {
357 LOG.debug("Cancelled transaction {} was head of the batch, resuming processing", tx);
361 // There are multiple frontend transactions in this batch. We have to report them as failed, which dooms this
362 // transaction chain, too. Since we just came off of a locked transaction, we do not have a ready transaction
363 // at the moment, but there may be some transaction in-flight. So we proceed to shutdown the backend chain
364 // and mark the fact that we should be turning its completion into a failure.
365 deadTx = new SimpleImmutableEntry<>(tx,
366 new CancellationException("Transaction " + frontendTx + " canceled").fillInStackTrace());
371 public synchronized void close() {
372 final PingPongTransaction notLocked = lockedTx;
373 Preconditions.checkState(notLocked == null, "Attempted to close chain with outstanding transaction %s",
376 // This is not reliable, but if we observe it to be null and the process has already completed,
377 // the backend transaction chain will throw the appropriate error.
378 Preconditions.checkState(shutdownTx == null, "Attempted to close an already-closed chain");
380 // This may be a reaction to our failure callback, in that case the backend is already shutdown
381 if (deadTx != null) {
382 LOG.debug("Delegate {} is already closed due to failure {}", delegate, deadTx);
386 // Force allocations on slow path, picking up a potentially-outstanding transaction
387 final PingPongTransaction tx = READY_UPDATER.getAndSet(this, null);
390 // We have one more transaction, which needs to be processed somewhere. If we do not
391 // a transaction in-flight, we need to push it down ourselves.
392 // If there is an in-flight transaction we will schedule this last one into a dedicated
393 // slot. Allocation slow path will check its presence and fail, the in-flight path will
394 // pick it up, submit and immediately close the chain.
395 if (inflightTx == null) {
396 processTransaction(tx);
402 // Nothing outstanding, we can safely shutdown
408 public DOMDataReadOnlyTransaction newReadOnlyTransaction() {
409 final PingPongTransaction tx = allocateTransaction();
411 return new DOMDataReadOnlyTransaction() {
413 public CheckedFuture<Optional<NormalizedNode<?, ?>>, ReadFailedException> read(
414 final LogicalDatastoreType store, final YangInstanceIdentifier path) {
415 return tx.getTransaction().read(store, path);
419 public CheckedFuture<Boolean, ReadFailedException> exists(final LogicalDatastoreType store,
420 final YangInstanceIdentifier path) {
421 return tx.getTransaction().exists(store, path);
425 public Object getIdentifier() {
426 return tx.getTransaction().getIdentifier();
430 public void close() {
431 readyTransaction(tx);
437 public DOMDataReadWriteTransaction newReadWriteTransaction() {
438 final PingPongTransaction tx = allocateTransaction();
439 final DOMDataReadWriteTransaction ret = new ForwardingDOMDataReadWriteTransaction() {
440 private boolean isOpen = true;
443 protected DOMDataReadWriteTransaction delegate() {
444 return tx.getTransaction();
448 public CheckedFuture<Void, TransactionCommitFailedException> submit() {
449 readyTransaction(tx);
451 return tx.getSubmitFuture();
456 public ListenableFuture<RpcResult<TransactionStatus>> commit() {
457 readyTransaction(tx);
459 return tx.getCommitFuture();
463 public boolean cancel() {
465 cancelTransaction(tx, this);
474 tx.recordFrontendTransaction(ret);
479 public DOMDataWriteTransaction newWriteOnlyTransaction() {
480 return newReadWriteTransaction();