package org.opendaylight.openflowplugin.openflow.md.queue;
import java.util.Collection;
-import java.util.Comparator;
import java.util.List;
import java.util.Map;
import java.util.concurrent.ArrayBlockingQueue;
import java.util.concurrent.BlockingQueue;
-import java.util.concurrent.ConcurrentSkipListSet;
+import java.util.concurrent.CopyOnWriteArraySet;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Future;
import java.util.concurrent.RejectedExecutionHandler;
import java.util.concurrent.ThreadPoolExecutor;
import java.util.concurrent.TimeUnit;
-
-import org.opendaylight.openflowplugin.openflow.md.core.IMDMessageTranslator;
+import org.opendaylight.openflowplugin.api.openflow.md.core.IMDMessageTranslator;
+import org.opendaylight.openflowplugin.api.openflow.md.core.TranslatorKey;
+import org.opendaylight.openflowplugin.api.openflow.md.queue.HarvesterHandle;
+import org.opendaylight.openflowplugin.api.openflow.md.queue.PopListener;
+import org.opendaylight.openflowplugin.api.openflow.md.queue.QueueItem;
+import org.opendaylight.openflowplugin.api.openflow.md.queue.QueueKeeper;
+import org.opendaylight.openflowplugin.api.openflow.md.queue.QueueProcessor;
+import org.opendaylight.openflowplugin.api.openflow.statistics.MessageSpy;
+import org.opendaylight.openflowplugin.api.openflow.statistics.MessageSpy.StatisticsGroup;
import org.opendaylight.openflowplugin.openflow.md.core.ThreadPoolLoggingExecutor;
-import org.opendaylight.openflowplugin.openflow.md.core.TranslatorKey;
-import org.opendaylight.openflowplugin.openflow.md.queue.MessageSpy.STATISTIC_GROUP;
import org.opendaylight.yang.gen.v1.urn.opendaylight.openflow.protocol.rev130731.OfHeader;
import org.opendaylight.yangtools.yang.binding.DataContainer;
import org.opendaylight.yangtools.yang.binding.DataObject;
/**
- * {@link QueueKeeper} implementation focused to keep order and use up mutiple threads for translation phase.
- * <br/>
- * There is internal thread pool of limited size ({@link QueueProcessorLightImpl#setProcessingPoolSize(int)})
+ * {@link org.opendaylight.openflowplugin.api.openflow.md.queue.QueueKeeper} implementation focused to keep order and use up mutiple threads for translation phase.
+ * <br>
+ * There is internal thread pool of limited size ({@link QueueProcessorLightImpl#setProcessingPoolSize(int)})
* dedicated to translation. Then there is singleThreadPool dedicated to publishing (via popListeners)
- * <br/>
+ * <br>
* Workflow:
* <ol>
* <li>upon message push ticket is created and enqueued</li>
* <li>available threads from internal pool translate the massage wrapped in ticket</li>
- * <li>when translation of particular message is finished, result is set in future result of wrapping ticket</br>
+ * <li>when translation of particular message is finished, result is set in future result of wrapping ticket<br>
* (order of tickets in queue is not touched during translate)
* </li>
* <li>at the end of queue there is {@link TicketFinisher} running in singleThreadPool and for each ticket it does:
* <li>invoke blocking {@link Future#get()} on the dequeued ticket</li>
* <li>as soon as the result of translation is available, appropriate popListener is invoked</li>
* </ol>
- * and this way the order of messages is preserved and also multiple threads are used by translating
+ * and this way the order of messages is preserved and also multiple threads are used by translating
* </li>
* </ol>
- *
- *
+ *
+ *
*/
public class QueueProcessorLightImpl implements QueueProcessor<OfHeader, DataObject> {
- protected static final Logger LOG = LoggerFactory
+ private static final Logger LOG = LoggerFactory
.getLogger(QueueProcessorLightImpl.class);
private BlockingQueue<TicketResult<DataObject>> ticketQueue;
private int processingPoolSize = 4;
private ExecutorService harvesterPool;
private ExecutorService finisherPool;
-
+
protected Map<Class<? extends DataObject>, Collection<PopListener<DataObject>>> popListenersMapping;
private Map<TranslatorKey, Collection<IMDMessageTranslator<OfHeader, List<DataObject>>>> translatorMapping;
private TicketProcessorFactory<OfHeader, DataObject> ticketProcessorFactory;
public void init() {
int ticketQueueCapacity = 1500;
ticketQueue = new ArrayBlockingQueue<>(ticketQueueCapacity);
- messageSources = new ConcurrentSkipListSet<>(
- new Comparator<QueueKeeper<OfHeader>>() {
- @Override
- public int compare(QueueKeeper<OfHeader> o1,
- QueueKeeper<OfHeader> o2) {
- return Integer.valueOf(o1.hashCode()).compareTo(o2.hashCode());
- }
- });
-
- processorPool = new ThreadPoolLoggingExecutor(processingPoolSize, processingPoolSize, 0,
- TimeUnit.MILLISECONDS,
- new ArrayBlockingQueue<Runnable>(ticketQueueCapacity),
+ /*
+ * TODO FIXME - DOES THIS REALLY NEED TO BE CONCURRENT? Can we figure out
+ * a better lifecycle? Why does this have to be a Set?
+ */
+ messageSources = new CopyOnWriteArraySet<>();
+
+ processorPool = new ThreadPoolLoggingExecutor(processingPoolSize, processingPoolSize, 0,
+ TimeUnit.MILLISECONDS,
+ new ArrayBlockingQueue<Runnable>(ticketQueueCapacity),
"OFmsgProcessor");
// force blocking when pool queue is full
processorPool.setRejectedExecutionHandler(new RejectedExecutionHandler() {
}
}
});
-
- harvesterPool = new ThreadPoolLoggingExecutor(1, 1, 0,
+
+ harvesterPool = new ThreadPoolLoggingExecutor(1, 1, 0,
TimeUnit.MILLISECONDS, new ArrayBlockingQueue<Runnable>(1), "OFmsgHarvester");
- finisherPool = new ThreadPoolLoggingExecutor(1, 1, 0,
+ finisherPool = new ThreadPoolLoggingExecutor(1, 1, 0,
TimeUnit.MILLISECONDS, new ArrayBlockingQueue<Runnable>(1), "OFmsgFinisher");
finisher = new TicketFinisherImpl(
ticketQueue, popListenersMapping);
finisherPool.execute(finisher);
-
+
harvester = new QueueKeeperHarvester<OfHeader>(this, messageSources);
harvesterPool.execute(harvester);
@Override
public void enqueueQueueItem(QueueItem<OfHeader> queueItem) {
- messageSpy.spyMessage(queueItem.getMessage(), STATISTIC_GROUP.FROM_SWITCH_ENQUEUED);
+ messageSpy.spyMessage(queueItem.getMessage(), StatisticsGroup.FROM_SWITCH_ENQUEUED);
TicketImpl<OfHeader, DataObject> ticket = new TicketImpl<>();
ticket.setConductor(queueItem.getConnectionConductor());
ticket.setMessage(queueItem.getMessage());
ticket.setQueueType(queueItem.getQueueType());
-
+
LOG.trace("ticket scheduling: {}, ticket: {}",
- queueItem.getMessage().getImplementedInterface().getSimpleName(),
+ queueItem.getMessage().getImplementedInterface().getSimpleName(),
System.identityHashCode(queueItem));
scheduleTicket(ticket);
}
-
-
+
+
@Override
public void directProcessQueueItem(QueueItem<OfHeader> queueItem) {
- messageSpy.spyMessage(queueItem.getMessage(), STATISTIC_GROUP.FROM_SWITCH_ENQUEUED);
+ messageSpy.spyMessage(queueItem.getMessage(), StatisticsGroup.FROM_SWITCH_ENQUEUED);
TicketImpl<OfHeader, DataObject> ticket = new TicketImpl<>();
ticket.setConductor(queueItem.getConnectionConductor());
ticket.setMessage(queueItem.getMessage());
-
+
LOG.debug("ticket scheduling: {}, ticket: {}",
- queueItem.getMessage().getImplementedInterface().getSimpleName(),
+ queueItem.getMessage().getImplementedInterface().getSimpleName(),
System.identityHashCode(queueItem));
-
+
ticketProcessorFactory.createProcessor(ticket).run();
-
+
// publish notification
finisher.firePopNotification(ticket.getDirectResult());
}
if (! added) {
LOG.debug("registration of message source queue failed - already registered");
}
- MessageSourcePollRegistration<QueueKeeper<OfHeader>> queuePollRegistration =
+ MessageSourcePollRegistration<QueueKeeper<OfHeader>> queuePollRegistration =
new MessageSourcePollRegistration<>(this, queue);
return queuePollRegistration;
}
-
+
@Override
public boolean unregisterMessageSource(QueueKeeper<OfHeader> queue) {
return messageSources.remove(queue);
}
-
+
@Override
public Collection<QueueKeeper<OfHeader>> getMessageSources() {
return messageSources;
}
-
+
@Override
public HarvesterHandle getHarvesterHandle() {
return harvester;
Code Review / openflowplugin.git / blobdiff