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.common.api.data;
10 import org.opendaylight.controller.md.sal.common.api.TransactionStatus;
11 import org.opendaylight.yangtools.concepts.Path;
12 import org.opendaylight.yangtools.yang.common.RpcResult;
14 import com.google.common.util.concurrent.CheckedFuture;
15 import com.google.common.util.concurrent.ListenableFuture;
18 * Write transaction provides mutation capabilities for a data tree.
21 * Initial state of write transaction is a stable snapshot of the current data tree.
22 * The state is captured when the transaction is created and its state and underlying
23 * data tree are not affected by other concurrently running transactions.
25 * Write transactions are isolated from other concurrent write transactions. All
26 * writes are local to the transaction and represent only a proposal of state
27 * change for the data tree and it is not visible to any other concurrently running
30 * Applications make changes to the local data tree in the transaction by via the
31 * <b>put</b>, <b>merge</b>, and <b>delete</b> operations.
33 * <h2>Put operation</h2>
34 * Stores a piece of data at a specified path. This acts as an add / replace
35 * operation, which is to say that whole subtree will be replaced by the
38 * Performing the following put operations:
41 * 1) container { list [ a ] }
42 * 2) container { list [ b ] }
45 * will result in the following data being present:
48 * container { list [ b ] }
50 * <h2>Merge operation</h2>
51 * Merges a piece of data with the existing data at a specified path. Any pre-existing data
52 * which is not explicitly overwritten will be preserved. This means that if you store a container,
53 * its child lists will be merged.
55 * Performing the following merge operations:
58 * 1) container { list [ a ] }
59 * 2) container { list [ b ] }
62 * will result in the following data being present:
65 * container { list [ a, b ] }
68 * This also means that storing the container will preserve any
69 * augmentations which have been attached to it.
71 * <h2>Delete operation</h2>
72 * Removes a piece of data from a specified path.
74 * After applying changes to the local data tree, applications publish the changes proposed in the
75 * transaction by calling {@link #submit} on the transaction. This seals the transaction
76 * (preventing any further writes using this transaction) and submits it to be
77 * processed and applied to global conceptual data tree.
79 * The transaction commit may fail due to a concurrent transaction modifying and committing data in
80 * an incompatible way. See {@link #submit} for more concrete commit failure examples.
82 * <b>Implementation Note:</b> This interface is not intended to be implemented
83 * by users of MD-SAL, but only to be consumed by them.
86 * Type of path (subtree identifier), which represents location in
89 * Type of data (payload), which represents data payload
91 public interface AsyncWriteTransaction<P extends Path<P>, D> extends AsyncTransaction<P, D> {
93 * Cancels the transaction.
95 * Transactions can only be cancelled if it's status is
96 * {@link TransactionStatus#NEW} or {@link TransactionStatus#SUBMITED}
98 * Invoking cancel() on {@link TransactionStatus#FAILED} or
99 * {@link TransactionStatus#CANCELED} will have no effect, and transaction
100 * is considered cancelled.
102 * Invoking cancel() on finished transaction (future returned by {@link #submit()}
103 * already completed with {@link TransactionStatus#COMMITED}) will always
104 * fail (return false).
106 * @return <tt>false</tt> if the task could not be cancelled,
107 * typically because it has already completed normally;
108 * <tt>true</tt> otherwise
114 * Removes a piece of data from specified path. This operation does not fail
115 * if the specified path does not exist.
118 * Logical data store which should be modified
121 * @throws IllegalStateException
122 * if the transaction is no longer {@link TransactionStatus#NEW}
124 void delete(LogicalDatastoreType store, P path);
127 * Submits this transaction to be asynchronously applied to update the logical data tree.
128 * The returned CheckedFuture conveys the result of applying the data changes.
130 * <b>Note:</b> It is strongly recommended to process the CheckedFuture result in an asynchronous
131 * manner rather than using the blocking get() method. See example usage below.
133 * This call logically seals the transaction, which prevents the client from
134 * further changing data tree using this transaction. Any subsequent calls to
135 * {@link #delete(LogicalDatastoreType, Path)} will fail with
136 * {@link IllegalStateException}.
138 * The transaction is marked as {@link TransactionStatus#SUBMITED} and
139 * enqueued into the data store back-end for processing.
142 * Whether or not the commit is successful is determined by versioning
143 * of the data tree and validation of registered commit participants
144 * ({@link AsyncConfigurationCommitHandler})
145 * if the transaction changes the data tree.
147 * The effects of a successful commit of data depends on data change listeners
148 * ({@link AsyncDataChangeListener}) and commit participants
149 * ({@link AsyncConfigurationCommitHandler}) that are registered with the data broker.
151 * <h3>Example usage:</h3>
153 * private void doWrite( final int tries ) {
154 * WriteTransaction writeTx = dataBroker.newWriteOnlyTransaction();
156 * MyDataObject data = ...;
157 * InstanceIdentifier<MyDataObject> path = ...;
158 * writeTx.put( LogicalDatastoreType.OPERATIONAL, path, data );
160 * Futures.addCallback( writeTx.submit(), new FutureCallback<Void>() {
161 * public void onSuccess( Void result ) {
165 * public void onFailure( Throwable t ) {
166 * if( t instanceof OptimisticLockFailedException ) {
167 * if( ( tries - 1 ) < 0 ) {
169 * doWrite( tries - 1 );
174 * // failed due to another type of TransactionCommitFailedException.
181 * <h2>Failure scenarios</h2>
183 * Transaction may fail because of multiple reasons, such as
185 * <li>Another transaction finished earlier and modified the same node in a
186 * non-compatible way (see below). In this case the returned future will fail with an
187 * {@link OptimisticLockFailedException}. It is the responsibility of the
188 * caller to create a new transaction and submit the same modification again in
189 * order to update data tree. <i><b>Warning</b>: In most cases, retrying after an
190 * OptimisticLockFailedException will result in a high probability of success.
191 * However, there are scenarios, albeit unusual, where any number of retries will
192 * not succeed. Therefore it is strongly recommended to limit the number of retries (2 or 3)
193 * to avoid an endless loop.</i>
195 * <li>Data change introduced by this transaction did not pass validation by
196 * commit handlers or data was incorrectly structured. Returned future will
197 * fail with a {@link DataValidationFailedException}. User should not retry to
198 * create new transaction with same data, since it probably will fail again.
202 * <h3>Change compatibility</h3>
204 * There are several sets of changes which could be considered incompatible
205 * between two transactions which are derived from same initial state.
206 * Rules for conflict detection applies recursively for each subtree
209 * <h4>Change compatibility of leafs, leaf-list items</h4>
211 * Following table shows state changes and failures between two concurrent transactions,
212 * which are based on same initial state, Tx 1 completes successfully
213 * before Tx 2 is submitted.
216 * <tr><th>Initial state</th><th>Tx 1</th><th>Tx 2</th><th>Result</th></tr>
217 * <tr><td>Empty</td><td>put(A,1)</td><td>put(A,2)</td><td>Tx 2 will fail, state is A=1</td></tr>
218 * <tr><td>Empty</td><td>put(A,1)</td><td>merge(A,2)</td><td>A=2</td></tr>
220 * <tr><td>Empty</td><td>merge(A,1)</td><td>put(A,2)</td><td>Tx 2 will fail, state is A=1</td></tr>
221 * <tr><td>Empty</td><td>merge(A,1)</td><td>merge(A,2)</td><td>A=2</td></tr>
224 * <tr><td>A=0</td><td>put(A,1)</td><td>put(A,2)</td><td>Tx 2 will fail, A=1</td></tr>
225 * <tr><td>A=0</td><td>put(A,1)</td><td>merge(A,2)</td><td>A=2</td></tr>
226 * <tr><td>A=0</td><td>merge(A,1)</td><td>put(A,2)</td><td>Tx 2 will fail, A=1</td></tr>
227 * <tr><td>A=0</td><td>merge(A,1)</td><td>merge(A,2)</td><td>A=2</td></tr>
229 * <tr><td>A=0</td><td>delete(A)</td><td>put(A,2)</td><td>Tx 2 will fail, A does not exists</td></tr>
230 * <tr><td>A=0</td><td>delete(A)</td><td>merge(A,2)</td><td>A=2</td></tr>
233 * <h4>Change compatibility of subtrees</h4>
235 * Following table shows state changes and failures between two concurrent transactions,
236 * which are based on same initial state, Tx 1 completes successfully
237 * before Tx 2 is submitted.
240 * <tr><th>Initial state</th><th>Tx 1</th><th>Tx 2</th><th>Result</th></tr>
242 * <tr><td>Empty</td><td>put(TOP,[])</td><td>put(TOP,[])</td><td>Tx 2 will fail, state is TOP=[]</td></tr>
243 * <tr><td>Empty</td><td>put(TOP,[])</td><td>merge(TOP,[])</td><td>TOP=[]</td></tr>
245 * <tr><td>Empty</td><td>put(TOP,[FOO=1])</td><td>put(TOP,[BAR=1])</td><td>Tx 2 will fail, state is TOP=[FOO=1]</td></tr>
246 * <tr><td>Empty</td><td>put(TOP,[FOO=1])</td><td>merge(TOP,[BAR=1])</td><td>TOP=[FOO=1,BAR=1]</td></tr>
248 * <tr><td>Empty</td><td>merge(TOP,[FOO=1])</td><td>put(TOP,[BAR=1])</td><td>Tx 2 will fail, state is TOP=[FOO=1]</td></tr>
249 * <tr><td>Empty</td><td>merge(TOP,[FOO=1])</td><td>merge(TOP,[BAR=1])</td><td>TOP=[FOO=1,BAR=1]</td></tr>
251 * <tr><td>TOP=[]</td><td>put(TOP,[FOO=1])</td><td>put(TOP,[BAR=1])</td><td>Tx 2 will fail, state is TOP=[FOO=1]</td></tr>
252 * <tr><td>TOP=[]</td><td>put(TOP,[FOO=1])</td><td>merge(TOP,[BAR=1])</td><td>state is TOP=[FOO=1,BAR=1]</td></tr>
253 * <tr><td>TOP=[]</td><td>merge(TOP,[FOO=1])</td><td>put(TOP,[BAR=1])</td><td>Tx 2 will fail, state is TOP=[FOO=1]</td></tr>
254 * <tr><td>TOP=[]</td><td>merge(TOP,[FOO=1])</td><td>merge(TOP,[BAR=1])</td><td>state is TOP=[FOO=1,BAR=1]</td></tr>
255 * <tr><td>TOP=[]</td><td>delete(TOP)</td><td>put(TOP,[BAR=1])</td><td>Tx 2 will fail, state is empty store</td></tr>
256 * <tr><td>TOP=[]</td><td>delete(TOP)</td><td>merge(TOP,[BAR=1])</td><td>state is TOP=[BAR=1]</td></tr>
258 * <tr><td>TOP=[]</td><td>put(TOP/FOO,1)</td><td>put(TOP/BAR,1])</td><td>state is TOP=[FOO=1,BAR=1]</td></tr>
259 * <tr><td>TOP=[]</td><td>put(TOP/FOO,1)</td><td>merge(TOP/BAR,1)</td><td>state is TOP=[FOO=1,BAR=1]</td></tr>
260 * <tr><td>TOP=[]</td><td>merge(TOP/FOO,1)</td><td>put(TOP/BAR,1)</td><td>state is TOP=[FOO=1,BAR=1]</td></tr>
261 * <tr><td>TOP=[]</td><td>merge(TOP/FOO,1)</td><td>merge(TOP/BAR,1)</td><td>state is TOP=[FOO=1,BAR=1]</td></tr>
262 * <tr><td>TOP=[]</td><td>delete(TOP)</td><td>put(TOP/BAR,1)</td><td>Tx 2 will fail, state is empty store</td></tr>
263 * <tr><td>TOP=[]</td><td>delete(TOP)</td><td>merge(TOP/BAR,1]</td><td>Tx 2 will fail, state is empty store</td></tr>
265 * <tr><td>TOP=[FOO=1]</td><td>put(TOP/FOO,2)</td><td>put(TOP/BAR,1)</td><td>state is TOP=[FOO=2,BAR=1]</td></tr>
266 * <tr><td>TOP=[FOO=1]</td><td>put(TOP/FOO,2)</td><td>merge(TOP/BAR,1)</td><td>state is TOP=[FOO=2,BAR=1]</td></tr>
267 * <tr><td>TOP=[FOO=1]</td><td>merge(TOP/FOO,2)</td><td>put(TOP/BAR,1)</td><td>state is TOP=[FOO=2,BAR=1]</td></tr>
268 * <tr><td>TOP=[FOO=1]</td><td>merge(TOP/FOO,2)</td><td>merge(TOP/BAR,1)</td><td>state is TOP=[FOO=2,BAR=1]</td></tr>
269 * <tr><td>TOP=[FOO=1]</td><td>delete(TOP/FOO)</td><td>put(TOP/BAR,1)</td><td>state is TOP=[BAR=1]</td></tr>
270 * <tr><td>TOP=[FOO=1]</td><td>delete(TOP/FOO)</td><td>merge(TOP/BAR,1]</td><td>state is TOP=[BAR=1]</td></tr>
274 * <h3>Examples of failure scenarios</h3>
276 * <h4>Conflict of two transactions</h4>
278 * This example illustrates two concurrent transactions, which derived from
279 * same initial state of data tree and proposes conflicting modifications.
282 * txA = broker.newWriteTransaction(); // allocates new transaction, data tree is empty
283 * txB = broker.newWriteTransaction(); // allocates new transaction, data tree is empty
285 * txA.put(CONFIGURATION, PATH, A); // writes to PATH value A
286 * txB.put(CONFIGURATION, PATH, B) // writes to PATH value B
288 * ListenableFuture futureA = txA.submit(); // transaction A is sealed and submitted
289 * ListenebleFuture futureB = txB.submit(); // transaction B is sealed and submitted
292 * Commit of transaction A will be processed asynchronously and data tree
293 * will be updated to contain value <code>A</code> for <code>PATH</code>.
294 * Returned {@link ListenableFuture} will successfully complete once
295 * state is applied to data tree.
297 * Commit of Transaction B will fail, because previous transaction also
298 * modified path in a concurrent way. The state introduced by transaction B
299 * will not be applied. Returned {@link ListenableFuture} object will fail
300 * with {@link OptimisticLockFailedException} exception, which indicates to
301 * client that concurrent transaction prevented the submitted transaction from being
304 * @return a CheckFuture containing the result of the commit. The Future blocks until the
305 * commit operation is complete. A successful commit returns nothing. On failure,
306 * the Future will fail with a {@link TransactionCommitFailedException} or an exception
307 * derived from TransactionCommitFailedException.
309 * @throws IllegalStateException
310 * if the transaction is not {@link TransactionStatus#NEW}
312 CheckedFuture<Void,TransactionCommitFailedException> submit();
315 * @deprecated Use {@link #submit()} instead.
318 ListenableFuture<RpcResult<TransactionStatus>> commit();