3. Release UMS 5.3.5 / UMP 5.3.5 / UMQ 5.3.5 - November - 2013

3.1.1. Updates

• The unicast resolver daemon, lbmrd, now has command-line options to set buffer size in bytes for the receive socket buffer (-r flag) and the send socket buffer (-s flag).

• When an error occurs on a socket while sending an ACK in LBT-RU, UM now generates an EOS immediately.

• The following error log messages are added to the Operations Guide:

  • Core-5688-3375: unicast resolver %s:%u went inactive.

  • Core-5688-3387: LBMR Extended Type 0x0 incorrect (%s:%d len %d). [%s]. Dropping.

  • CoreApi-5688-3287: could not find open unicast source port in range [%d-%d]

  • CoreApi-5688-3320: lbm_socket_recv recv/recvfrom: %s

• In the .NET API, UM now re-uses LBMChannelInfo, UMQMessageId, and UMQIndexInfo objects between different LBMMessage objects for delivery, reducing garbage collection overhead.

• The lbmrd resolver daemon now uses blocking sends for all outgoing data.

3.1.2. Bug Fixes

• In the UM Gateway, single TCP peer connections under Windows now function correctly.

• UM no longer returns -1 from lbm_context_process_events() for non-critical errors.

• Corrected a problem with how UM handles connect failures under Windows, which was causing the UM Gateway to not reconnect with other UM Gateway peer portals after a timeout condition.

• Corrected a problem where a fatal assert would occur when a receiver received a message containing a TSNI request. This could happen when a receiver late joined from an older-version source.

3.1.3. Known Issues

• When calling lbm_src_send() from the context thread inside a lbm_timer callback, a source can freeze when it hits rate limiter limits, requiring an application restart.

3.1.4. Version Compatibilities

• LBT-IPC, when used on UM 5.3.x, is not compatible with LBT-IPC used on UM 5.2.x or earlier.

3.2.1. Updates

• In the Ultra Messaging Guide for Persistence, the following sections have been changed or added:

  • 2.1.5 Message Stability

• The following error log message is updated in the Operations Guide:

  • Core-5688-542: received ACK for unknown source from %s:%d\n

3.3.1. Updates

• In the Ultra Messaging Guide for Persistence and Queuing, the following sections have been changed or added:

  • 1.2 Queuing

  • 1.3 Semantic Responsibilities

  • 2.2.2 Message Stability

  • 2.2.3 Once-and-Only-Once Delivery

  • 2.2.8 Queue Fault Tolerance

  • 2.2.9 Indexed Queuing

If you are considering queuing semantics for your application, please contact Informatica.

• In the Ultra Messaging JMS Guide the following sections have been changed:

  • 1. Introduction, known issues.

3.3.2. Known Issues

• Configurations with multiple queue instances (slaves) can lead to inconsistent states, which can cause message loss, crashes or an inability to restart without removal of files (also resulting in message loss). For guidance, see Section 2.2.8 of the Ultra Messaging Guide for Persistence and Queuing.

• A queue instance may crash with the following log message if allow-queue-browsing is set to 1: [EMERGENCY]: FATAL: failed assertion [sinc_q_msg_info_node!=NULL] at line 5998 in ../../../../src/stored/umqueue.c.

• UMQ flags reassigned messages upon delivery, but when a queue restarts, UMQ may re-deliver some messages without flagging them.

• Queues may occasionally crash upon restart if the queue option forwarding-behavior is set to store-while-forward. Set this option to store-then-forward.

• A queue instance may crash with the following log message if messages are of 0 length: [EMERGENCY]: FATAL: failed assertion [msg->lbm_msg!=NULL] at line 3896 in ../../../../src/stored/umqsinc.c