Description:

•    TIBCO Enterprise Message Service 4.4.1 Hotfix8 has been released.

•    This Hotfix8 is cumulative for the components listed below. It includes previously generated TIBCO Enterprise Message Service 4.4.1 hotfixes.

•    TIBCO Enterprise Message Service 4.4.1 Hotfix7 does not contain TIBCO Enterprise Message Service 4.4.1 Hotfix4 defect corrections; this Hotfix8 corrects this omission and fixes several problems in addition to that.


======================================================================
Affected Files for 4.4.1-hotfix08

This hotfix is available on the following platforms:
    linux24gl23/x86
    linux24gl23/ia64
    sol27/sparc
    sol210/x86
    hpux110/hppa
    hpux112/ia64
    win32/x86/vc8
    win64/x86
    aix51/power
    darwin/power
    darwin/x86
    tru6451/alpha

Affected files:
    tibemsd and tibemsd64

======================================================================
Closed Issues in 4.4.1-hotfix08 (this release):

  Defect ID: 1-8EYRZV and 1-8EYS0R
     If the export property of a wildcard topic was changed, and a
     bridge topic inherited the export property, the server did not
     notify the other servers. This has been fixed.

  Defect ID: 1-8ASFJ4
     The server sent an unnecessary protocol message when the list of
     subscribers to a destination bridge topic changed. This has been
     fixed.

  Defect ID: 1-84JMOT
     The server sent an unnecessary protocol message when it received
     a routed message, if consumers were subscribed to the topic but
     the message did not match the consumers' selectors. This has been
     fixed.


======================================================================
Closed Issues in 4.4.1-hotfix06:

  The consumed_msg_hold_time flag, which was removed in the EMS 4.4
release, has been restored for backward compatibility.

  Defect ID: 1-866WO5
     Durable subscribers failed when the connection was restarted immediately after a
previous connection closed. This has been fixed.

  Defect ID: 1-85NQ4I
     When a durable subscriber was created using the administration tool, with the properties
clientid and selector, the clientid property was not stored in the durables.conf file.
On EMS server restart, duplicate durable subscribers were created. This has been fixed.

  Defect ID: 1-8CQMCB
     The server or C library could crash if TX tracing was enabled
     when it processed an XID with a large number of non-printable
     characters. This has been fixed.

  Defect ID: 1-8CQWNO
     The tibemsadmin tool could not commit or rollback an XA transaction
     if the XID had a large number of non-printable characters.  
     This has been fixed.

  Defect ID: 1-8B0VFG
     The EMS server would core dump in a rare condition related to
     route when a flow stall occurred in the receiving EMS server.
     This has been fixed.

======================================================================
Closed Issues in 4.4.1-hotfix04:

  Defect ID: 1-85GCU1
     Fixed an error that prevented the removal of a temporary queue,
     created as a reply-to destination between routing partners, after
     the sender and receiver connections are destroyed. Queues are now
     removed at regular 3-minute intervals.

  Defect ID: 1-873JVN
     Messages that were published to a dynamic queue and that were
     part of an XA transaction and in a XA prepared state when
     recovered by the EMS server, were not being delivered to queue
     consumers even after a successful commit of the XA transaction.
     This has been fixed.

  Defect ID: 1-885OZV
     Messages that were received by a queue consumer as part of an XA
     transaction, and which were recovered in an XA prepared state
     during an EMS server restart, did not appear in the dynamic
     queue. This has been fixed.


======================================================================
Closed Issues in 4.4.1-hotfix03:

  Defect ID: 1-84ZA3C
     Fixed a defect that caused the EMS Server to stall and not resume
     when a server processing incoming messages reached memory limits.
     Flow stall when a downstream server reaches memory limits is
     expected behavior. However, message flow should resume after the
     error condition is cleared, such as when messages are purged and
     memory usage drops below the limit.


======================================================================
Closed Issues in 4.4.1-hotfix02:

  Defect ID: 1-838PFR
     A change in Release 4.4.0 resulted in a performance issue with
     the data store for valid messages. This has been fixed.

  Defect ID: 1-7W1MX1
     The error "Error: Global or routed queue cannot specify
     prefetch=none" was returned when using "addprop" to set any
     property. This has been fixed so that properties (except for
     prefetch=none) can now be set on global or routed queues.
    
  Defect ID: 1-837ZMF
     If an application that uses a client ID disconnects or terminates
     before a commit() succeeds, immediately reconnecting was not
     always successful. This has been fixed.

  Defect ID: 1-81ZZCH
     Fixed a defect that could cause the server to not recover valid
     messages if, prior to the restart, the server ran out of memory
     at a very specific point in time while creating an internal
     record that is later persisted on disk. This memory failure may
     not be repored if the user is not using reserve_memory. Of
     course, users should always ensure that the EMS Server has
     appropriate resources. You can limit the memory usage by setting
     the max_msg_memory parameter according to the system's memory
     availability.

======================================================================
Closed Issues in 4.4.1-hotfix01:

  Defect ID: 1-82HYZW
     In EMS 4.4.0, messages in the $sys.undelivered queue that are
     consumed are not removed from the "Pending Message" count.
     This has been fixed.

  Defect ID: 1-820SBH
    Fixed a defect that could cause the EMS Server to crash when it
    would try to discard NON_PERSISTENT or RELIABLE messages that
    have reached their maximum delivery count based on the queue's
    maxRedelivery property. Messages with the
    TIBCO_PRESERVE_UNDELIVERED boolean property would be unaffected
    and are correctly moved to the $sys.undelivered queue.      

  Defect ID: 1-81UJTJ
    Under an operational mode that disables batched synchronous
    writes, the server may return the debug message "DEBUG:
    Insufficient buffer - SEVERE ERROR: Failed writing message".  
    This has been fixed.
  
=======================================================================

Installation

    Stop the applications and EMS server.
    Replace the EMS server binary with the new version supplied.
    Start the EMS server and its applications.

======================================================================


* You can download this 4.4.1 Hotfix8 from the TIBCO Product Support FTP server using your eSupport username and password, at

            ftp://support-ftp.tibco.com/available_downloads/ActiveEnterprise/Messaging/EMS/4.4.1/