Mailing-List: contact users-help@activemq.apache.org; run by ezmlm
Precedence: bulk
Reply-To: users@activemq.apache.org
Received-SPF: neutral (athena.apache.org: 216.139.236.26 is neither permitted
 nor denied by domain of geeky.diamond@gmail.com)
Date: Mon, 6 Jun 2011 07:02:28 -0700 (PDT)
From: "geeky.diamond" <geeky.diamond@gmail.com>
To: users@activemq.apache.org
Message-ID: <1307368948781-3576980.post@n4.nabble.com>
Subject: XA transaction rollback memory leak
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

We have Glassfish sending messages to an activeMQ topic using an XA
Transaction. This seems to work correctly when transactions are committing. 
However, *every* time we have a transaction rollback we have a slight memory
leak from activeMQ. After 1000 rollbacks our 1mb queue memory is full. 
The logs show: 
2011-06-06 11:50:06,842 | DEBUG | Setting up new connection id:
ID:HOTCPC8567-53061-1307352319478-0:338, address: /127.0.0.1:56674 |
org.apache.activemq.broker.TransportConnection | ActiveMQ Transport:
tcp:///127.0.0.1:56674 
2011-06-06 11:50:06,843 | DEBUG | localhost adding consumer:
ID:HOTCPC8567-53061-1307352319478-0:338:-1:102 for destination:
topic://ActiveMQ.Advisory.TempQueue,topic://ActiveMQ.Advisory.TempTopic |
org.apache.activemq.broker.region.AbstractRegion | ActiveMQ Transport:
tcp:///127.0.0.1:56674 
2011-06-06 11:50:06,846 | DEBUG | XA Transaction new/begin :
XID:4871251:c51600003f724564484f54435043383536372c7365727665722c5033373030:484f54435043383536372c7365727665722c50333730302c00
| org.apache.activemq.transaction.XATransaction | ActiveMQ Transport:
tcp:///127.0.0.1:56674 
2011-06-06 11:50:06,849 | DEBUG |
Main:memory:topic://createBookingTopic:memory: usage change from: 9% of
available memory, to: 10% of available memory |
org.apache.activemq.usage.Usage | ActiveMQ Transport: tcp:///127.0.0.1:56674 
2011-06-06 11:50:06,852 | DEBUG | XA Transaction rollback:
XID:4871251:c51600003f724564484f54435043383536372c7365727665722c5033373030:484f54435043383536372c7365727665722c50333730302c00
| org.apache.activemq.transaction.XATransaction | ActiveMQ Transport:
tcp:///127.0.0.1:56674 
2011-06-06 11:50:06,884 | DEBUG | localhost removing consumer:
ID:HOTCPC8567-53061-1307352319478-0:338:-1:102 for destination:
topic://ActiveMQ.Advisory.TempQueue,topic://ActiveMQ.Advisory.TempTopic |
org.apache.activemq.broker.region.AbstractRegion | ActiveMQ Transport:
tcp:///127.0.0.1:56674 
2011-06-06 11:50:06,885 | DEBUG | remove connection id:
ID:HOTCPC8567-53061-1307352319478-0:338 |
org.apache.activemq.broker.TransportConnection | ActiveMQ Transport:
tcp:///127.0.0.1:56674 

So no obvious issue in ActiveMQ. Also, this does not happen when we use a
queue and the transaction is rolled back. 

This is a serious problem for us as we are needing to restart activeMQ on a
daily basis to keep the production servers running. Any help very much
appreciated. 

Tom

(Reposting so that the message goes to the mailing list)

--
View this message in context: http://activemq.2283324.n4.nabble.com/XA-transaction-rollback-memory-leak-tp3576980p3576980.html
Sent from the ActiveMQ - User mailing list archive at Nabble.com.