QPID-3280: Performance problem with TTL messages.

When sending a large number of messages with nonzero TTLs to a
cluster, overall message throughput drops by around 20-30% compared to
messages with TTL 0.

The previous approach to TTL in the cluster is replaced with a simpler
"cluster clock". Also QueueCleaner is executed in the cluster timer,
and modified to be deterministic in a cluster.

git-svn-id: https://svn.apache.org/repos/asf/qpid/trunk/qpid@1136170 13f79535-47bb-0310-9956-ffa450edef68

1 files changed, 16 insertions, 10 deletions

diff --git a/cpp/src/qpid/cluster/FailoverExchange.cpp b/cpp/src/qpid/cluster/FailoverExchange.cpp
index 84232dac1b..cfbe34a460 100644
--- a/cpp/src/qpid/cluster/FailoverExchange.cpp
+++ b/cpp/src/qpid/cluster/FailoverExchange.cpp
@@ -7,9 +7,9 @@
  * to you under the Apache License, Version 2.0 (the
  * "License"); you may not use this file except in compliance
  * with the License.  You may obtain a copy of the License at
- * 
+ *
  *   http://www.apache.org/licenses/LICENSE-2.0
- * 
+ *
  * Unless required by applicable law or agreed to in writing,
  * software distributed under the License is distributed on an
  * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
@@ -39,8 +39,10 @@ using namespace broker;
 using namespace framing;
 
 const string FailoverExchange::typeName("amq.failover");
-    
-FailoverExchange::FailoverExchange(management::Manageable* parent, Broker* b)  : Exchange(typeName, parent, b ) {
+
+FailoverExchange::FailoverExchange(management::Manageable* parent, Broker* b)
+    : Exchange(typeName, parent, b ), ready(false)
+{
     if (mgmtExchange != 0)
         mgmtExchange->set_type(typeName);
 }
@@ -53,16 +55,17 @@ void FailoverExchange::setUrls(const vector<Url>& u) {
 void FailoverExchange::updateUrls(const vector<Url>& u) {
     Lock l(lock);
     urls=u;
-    if (urls.empty()) return;
-    std::for_each(queues.begin(), queues.end(),
-                  boost::bind(&FailoverExchange::sendUpdate, this, _1));
+    if (ready && !urls.empty()) {
+        std::for_each(queues.begin(), queues.end(),
+                      boost::bind(&FailoverExchange::sendUpdate, this, _1));
+    }
 }
 
 string FailoverExchange::getType() const { return typeName; }
 
 bool FailoverExchange::bind(Queue::shared_ptr queue, const string&, const framing::FieldTable*) {
     Lock l(lock);
-    sendUpdate(queue);
+    if (ready) sendUpdate(queue);
     return queues.insert(queue).second;
 }
 
@@ -84,7 +87,7 @@ void FailoverExchange::sendUpdate(const Queue::shared_ptr& queue) {
     // Called with lock held.
     if (urls.empty()) return;
     framing::Array array(0x95);
-    for (Urls::const_iterator i = urls.begin(); i != urls.end(); ++i) 
+    for (Urls::const_iterator i = urls.begin(); i != urls.end(); ++i)
         array.add(boost::shared_ptr<Str16Value>(new Str16Value(i->str())));
     const ProtocolVersion v;
     boost::intrusive_ptr<Message> msg(new Message);
@@ -96,9 +99,12 @@ void FailoverExchange::sendUpdate(const Queue::shared_ptr& queue) {
     header.get<MessageProperties>(true)->getApplicationHeaders().setArray(typeName, array);
     AMQFrame headerFrame(header);
     headerFrame.setFirstSegment(false);
-    msg->getFrames().append(headerFrame);    
+    msg->getFrames().append(headerFrame);
     DeliverableMessage(msg).deliverTo(queue);
 }
 
+void FailoverExchange::setReady() {
+    ready = true;
+}
 
 }} // namespace cluster