Chapter 3. Fixed issues

ENTMQBR-1444 - [amqp] when receiver client connects without source being set, broker prints NPE

In previous releases, if an AMQP receiver client connected to the broker without the receiver’s source attribute (:source) being set, the broker displayed a "NullPointerException" warning. This issue is resolved. The broker no longer shows an exception if an AMQP receiver client does not have the :source attribute set when it connects to the broker.

ENTMQBR-2707 - [amqp] when receiver client connects without source being set, broker prints NPE

Previously, if AMQP messages in a pre-fetch buffer were returned to the queue, unacknowledged messages were returned in the wrong order. As a result of this issue, clients might later consume these messages in the wrong order, negatively affecting applications for which message ordering was important. This issue is resolved. Pre-fetched AMQP messages are always returned to the queue in the original order.

ENTMQBR-2767 - [AMQ7, purge message, OutOfMemoryException] with a large queue size, removeAllMessages() takes a long time and can eventually result in an OOM exception

Previously, for a queue with a large number (that is, in the order of millions) of messages, the JMX operation removeAllMessages could take up to several minutes to complete. In addition, the operation sometimes failed to complete, returning an out-of-memory (OOM) exception instead. This issue is now resolved.

ENTMQBR-2878 - AMQ broker does not clean the connection (MQTT) when the connection is broken

Previously, when an MQTT client dropped its connection to the broker without properly closing it, the broker did not close the underlying physical connection. This issue is now resolved.

ENTMQBR-2889 - Don’t delete auto-created queues when FORCE is used for configuration changes

Previously, if you specified config-delete-addresses=FORCE in your broker.xml configuration file, auto-created addresses and their associated queues were also deleted when the configuration was reloaded. This issue is now resolved.

ENTMQBR-2912 - Messages greater than 50kb do not appear in the AMQ Broker management console

Previously, if a large message was compressed and the compressed size fell below the minimum large message size, the message was sent as a normal message and did not appear in the AMQ Broker management console. This issue is now resolved. All large messages can be browsed in the management console, regardless of whether they have been compressed or not.

ENTMQBR-2932 - Remote JMX server on slave shuts down during failback

Previously, when a slave broker in a master-slave pair failed back to the original master broker, the JMX sever on the slave broker did not restart and could no longer be used. As a result of this issue, you could not manage or control the slave broker via JMX. However, other management methods (for example, via Jolokia or HTTP) were unaffected. This issue is now resolved. When a slave broker fails back to the original master broker, the JMX server is restarted.

ENTMQBR-3012 - [AMQ7, large messages] LargeMessage doesn’t make a full copy of its props

Previously, a copy of a large message still referenced some properties of the original message. This caused unexpected behaviour when a thread modified the properties. This issue is resolved. When the broker copies a large message, all properties are included in the copy.

ENTMQBR-3069 - AMQ JMS client doesn’t recover after a complete outage

Previously, if a master-slave broker pair failed back to the original master broker following an outage, an AMQ JMS client could not reconnect to the master. This failed connection was caused by failure of the broker to remove its existing client IDs following failback. This issue is now resolved.

ENTMQBR-3128 - Failback does not work for master/slave group using NFS shared store

In AMQ Broker 7.5, if you configured a live-backup broker group to use the shared store high availability (HA) policy with a Network File System (NFS), the backup broker did not fail back to the original live broker when the live broker became available again after a previous failure. This issue occurred even if you set allow-failback to true in the configuration of your backup broker. This issue is now resolved.

ENTMQBR-3136 - Improper Quoting in Generated artemis.profile File - Causing Start Failures in Some Environments

In previous releases, the JAVA_ARGS line of the generated artemis.profile file incorrectly included quotation marks around the default value of the hawtio.offline parameter (that is, Dhawtio.offline="true"). This formatting error could cause the broker to fail to start in some environments. This issue is now resolved.

ENTMQBR-3163 - Memory Leak when Opening and Closing AMQP Consumers in the Same Session / Context

Previously, if you consecutively opened and then closed a series of AMQP clients within the same JMSContext instance, this caused a memory leak on the broker. This issue is now resolved.

ENTMQBR-3215 - Broker rejects client re-connection after broker stop/start

Previously, if a broker was stopped and restarted, an AMQP client (for example, an AMQ JMS client) with reconnect logic could not reconnect to the broker. Specifically, the broker rejected the connection, indicating that the client ID already existed. This issue is now resolved.