Description:
The patch can be downloaded from the TIBCO Support ftp server, support-ftp.tibco.com. Please use your eSupport username and password to access the server. After connecting, go to available_downloads/ActiveEnterprise/InConcert/6.2.1/hotfix-6  to download the appropriate hotfix for your OS/platform.

The patched files are named based on platform and database, for instance, TIB_ic_6.2.1HF6_aix_o92.tar.gz for AIX with Oracle 9.2.x.


Note there is a readme, TIB_InConcert_6.2.1_Readme_hotfix06.txt, and a separate document for the database install, TIB_InConcert_6.2.1_DBinstall_hotfix06.txt.  Both can be downloaded separated but are in included in each .zip or .gz file.

HOTFIX6: Fixes in TIBCO InConcert 6.2.1 HF6 provide performance improvements in
contention in parallel operations of acquire/complete task and
jobCreateActive/Template calls. Additionally, support for RVCM for Notify
InOrder messages has been added.

This is a roll up release and contains all previous Hotfixes 1 through 5.

HF6 for DB2 Prerequisite: Must apply IBM DB2 patch7a (Version 8.2 (Version
8.1 FixPak 7 or higher) for AIX or Windows support.

======================================================================
Bug Fixes included in TIBCO InConcert 6.2.1 HF6:

Defects/SR's addressed:
----------------------

Defect Id: 1-25TX0B (SR ID: 1-24SW41)
Deadlock aborts happen 10% of time on job.createActive in MS SQL 2000.

Defect Id: 1-2J3G1P (SR ID: 1-26AZKD)
Bind option in DB 2 should be set to ANY so DB 2 can take advantage of INTRA
PARALLELISM.

Defect Id: 1-2T8M93 (SR ID: 1-28IZ9Z)
Deadlocks in DB2 Production.

Defect Id: 1-2T8M9K (SR ID: 1-1D0W6T)
Deadlocks with SQL Server 2000.

Defect Id: 1-2T8M9R (SR ID: 1-24SW41)
About 10% of workflow creation requests fail.

Bug Fixes in TIBCO InConcert 6.2.1HF2

Defect Id: 1-301RZ3 (SR ID: 1-2WX9V9)
Adding CACHE to * _ID_MAP can improve performance in the Oracle version of
InConcert.

Defect Id: 1-301RZL (SR ID: 1-2WX9V9)
Compressing and changing the Attribute index can improve performance in the
Oracle InConcert version.

Defect Id: 1-301SOO (SR ID: 1-2WX9V9)
Additional indexes which when compressed improve performance for InConcert using
Oracle.

Defect Id: 1-383GQO (SR ID: 1-1Y7CYD)
Frequent persistent table locks on InConcert database locking processing.

Details:
-------

1. Oracle/MSSQL/DB2 - Fix deadlocks on concurrent createActive & createTemplate
as well as acquire/complete. [NOTE: Sybase uses modified old method getting only
one task at a time. Expect error "noTaskReadyForPool" to be returned when there
might have previously been a deadlock. Though deadlocks should be reduced or
eliminated, the burden is on the client to retry until an acquired task is returned.
DB2 may also still see some deadlocks on acquire/complete though the number should
be reduced - this is highly variable.]

2. Oracle/MSSQL/Sybase/DB2 - Ensure equivalentRoles map does not cause deadlocks

3. Server - Fix SR ID 1-3CNUUX: Changed the structure of event message
published to include field name "inOrderMsg" so field name is not null

4. Windows Server Manager / Service change - Fixes [INCONCERT_SERVICE]
SetServiceStatus error

5. All platforms IcDirectoryDmn/IcDirMon - Fix for persistent search so that
changes come in order of LDAP operation

6. HPUX 11 / Oracle 9.2.0: Change includes use of try/catch block and new
IcServer:emem object used to hold 4M of memory in reserve for outOfMemory errors.
Added "-N" switch to build to allow 32 bit binary to grow past ~1G boundary and
to max ~2G of memory. The server was core dumping on outOfMemory errors on HPUX 11.

7. Oracle database changes include use of INITRANS, COMPRESS and
addition/removal of indices of TASKS table. In particular reversion to use of
TASKS_OVERDUE and TASKS_LATE_ST and dropping of 5 new indices introduced in HF2.

8. New in HF6: Undocumented environment variable to enable use of RVCM by
IcNotifier.
The following parameter can be set in the environment to indicate that RVCM
should be used by the IcNotifier:

IC_NOTIFIER_EVENT_PROCESSING_MODE

When the value of this parameter is set to "CERTIFIED" the IcNotifier will use a
ledger file located in the $IC_HOME/ledgers directory. The name of the ledger
file is &ltserver_name>_ne_cm.  The value of the parameter
IC_TIBRV_PUBLISH_EVENT_CM_TIMELIMIT, which can be specified using IcStart on
UNIX or the Server Manager on Windows, is used to specify the certified message
timeout. This parameter name is shared with the EAD and its use of CM for
PublishEvent messages. It is possible to use Reliable Publish Event messages
while using Certified Notifier messages.

A related parameter is the cmName which is a reusable name bound to the
TibrvCmTransport.

The IcNotifier CM transport has a name which is reusable. The name identifies
the CM transport to other CM transports, and is part of the CM label that
identifies outbound messages from the CM transport.

When a CM transport with a reusable name also has a file-based ledger, it
operates as an instance of a persistent correspondent which allows continuity of
certified delivery beyond transport invalidation and program restarts (for more
information, see Persistent Correspondents).

The name pattern used for the IcNotifier CM transport is: CM_&ltserver_name>_NE