The patch notes contain the
following sections:
This patch contains fixes for the
following CRs based on customer reported and internally discovered issues.
|
Problem tracking numbers Number |
As of Version |
Problem |
|
15262, 26462 |
4.1.0.8 |
Problem: Increasingly
slower response with repository access |
|
19930 |
4.1.0.8 |
Problem: the server CPU utilization goes as high as 95% to 100% when try to get DB
connections |
|
13902 |
4.1.0.8 |
Encrypt/Decrypt password causes lot of performance degradation. |
|
19420 |
4.1.0.8 |
This problem has been brought up
when an XML document containing empty tags was passed to the ICS. When the
document is sent back to a connector, the empty tags are no longer written to
the xml document. |
|
17342. |
4.1.0.8 |
Merge Scripts modification for
4.0.1 into 4.1.0. Also add vbroker.agent.addr into the CWSharedEnv.sh |
|
|
4.1.0.8 |
Fix the bug that it is
possible that connector controller start more than one thread for OAD restart
agent |
|
17785, 17831 |
4.1.0.7 |
A synchronization problem in the event sequencing was
causing multiple BOs to be generated in a group collaboration with two
collaborations A and B. Collaborations A's from port was bound to a source
connector, Collaboration B's to port was bound to a destination connector.
Collaboration A was making 2 service calls to Collaboration B. Collaboration
B was sending multiple events to the destination connector. It was detected
due to synchronization problem; the LockObjectEventId obtained from the
flowcontext was not valid. The fix involved storing and retrieving the
ObjectEventId in/from SequencedItem instead of depending on the flowcontext |
|
15917 |
4.1.0.7 |
Problem: In collaboration template, when an exception occurs, it could bubble up to a Main scenario of the collab. The flow of a collaboration could go to different pathes based on different messages embedded in the exception. Before release 4.0, we have no way of returning the message number from the exception object. Therefore, our customer uses hard-coded "Error" string to parse message number out of error message.
In 3.1.x version of ICS, the error message is in a format of: "Error # Messagecontent" (with a space after Error).
However, started with 4.x release of ICS, the format was changed to: "Error# Messagecontent" (without a space after Error).
When a customer upgrades to 4.x release of ICS, the collaboration templates in version 3.1.x stopped working if they use above logic and hard-coded to parse “Error #”. At the same time, customers who are using 4.x, and have already developed collaboration using the new format to parse message number.
Solution: To satisfy both 3.x and 4.x customers without forcing them to change their content, this patch changed the error format to: "Error # MessageContent Error#"
Impact on customer: None. The goal of this patch is to minimize user impact. Their content should work as before.
In the future: However, we do recommend customer to use CollaborationException's getMsgNumber() API to retrieve message number, if they need to use the message number to determine which path to choose.
For example, instead of doing: String errorPath = "Error 1234"; if (currentException.getMessage().indexOf(errorPath)>= 0) ...
customer can/should do: if (currentException.getMsgNumber() == 1234) ... |
|
16002 |
4.1.0.3 |
This CR fixes the limitation with SQL JDBC driver used with ICS 4.1.1. The SQL driver is not supporting retrieval of data more than 2KB in size. This problem was encountered in Pip3C3 process . |
|
16003 |
4.1.0.3 |
This CR fixes limitation
in the Oracle JDBC driver in ICS 4.1.1. The driver is not supporting
retrieving of data that exceeds 32KB in size. This problem was found in
Pip3A4 RNET process. |
|
15347
|
4.1.0.3
|
This CR fixes the situation where multiple BO's are
received from a collaboration group, when only a single BO is expected. |
|
15617 |
4.1.0.3 |
The problem fixed in this
CR is that once a <none > is explicitly bond to a BO supported by a connector,
then stop all running associated maps and stop ICS server. The server failed
to boot next time. |
|
16109
|
4.0.1.5 |
We can now stop the connectors from the
command line without OAD restarting them. The three methods to stop connectors are explained below:
a) connector_manager -kill - looks up the PID and kills the process
b) connector_manager -stop - communicates directly with the agent and
has the agent shut itself down.
c) connector_manager -stopgraceful - communicates with the ICS and has
the server shut down the agent, thereby letting OAD know not to restart (in
case of OAD auto restart feature is on).
|
|
15599 |
4.0.1.5
|
Modify UNIX ics_manager scripts to check if
the same ICS server has been already started. If so, the user cannot start
the second ICS server. |
|
15870 |
4.0.1.5
|
Modify UNIX connector_manager scripts to
check if the same connector agent has been already started. If so, the user cannot
start the second one. |
|
16027 |
4.0.1.4
|
MQ error when OAD tries to start the
connector agent in MQ native binding mode, such that connector agent start at
the pause state instead of active |
|
15951 |
4.0.1.3
|
ICS cannot address >= 2 GB of free memory |
|
15589 |
4.0.1.3
|
MQSeries 2053 (queue full) errors |
|
15949/15427 |
4.0.1.3
|
ics_manager -IsServerAlive false but CSM can
connect AND ics_manager -stop not working after changing ICS password |
|
15984 |
4.0.1.3
|
Visibrokers on different ports require system
environment variable set to save locally |
|
15334 |
4.0.1.3
|
Connector listeners getting 2009 MQSeries
errors and events get stuck in MQ until ICS reboot |
|
15347 |
4.0.1.3
|
Duplicate BOs created from collaboration
group |
|
15928 |
4.0.1.3
|
CSM statistics window shows Queue depth as
blank after catching 2009 |
|
15599 |
4.0.1.3
|
Second, mistaken ICS start causes logging to Arc_01.log
instead of Interchange.log when archive logging enabled |
|
15977 |
4.0.1.3 |
IDL:PortableServer/POA/WrongAdapter:1.0 error
in an HA configuration or on a machine that has mulithomed host machine (e.g. one physical IP address and one virtual
IP address). Specification of the IP address of ORB POA engine (OAHOST). A
new env variable, ORB_OAHOST has been introduced. Set this env is to the
particular IP address (e.g. set to thevirtual IP), to force the ICS server
and connector agents to only use that IP to export their IOR info, such that
we can avoid theconfusion between multihomed IP addresses. |
To perform an upgrade to this
version of the ICS:
|
Platform [Win32;Solaris;AIX] |
Add / Replace / Remove |
File (Starting from WebSphere BI install directory |
|
|
4.1.0.8 |
Win32:Solaris:AIX |
Replace |
lib/Collabutils.jar |
|
4.1.0.8 |
Win32:Solaris:AIX |
Replace |
lib/CrossWorlds.jar |
|
4.1.0.8 |
Win32:Solaris:AIX |
Replace |
lib/mq.jar |
|
4.1.0.8 |
Win32:Solaris:AIX |
Replace |
InterchangeSystem.txt |
|
4.1.0.8 |
Solaris:AIX |
Replace |
bin/connector_manager |
|
4.1.0.8 |
Solaris:AIX |
Replace |
bin/CWSharedEnv.sh |
|
4.1.0.8 |
Solaris:AIX |
Replace |
bin/ics_manager |
|
4.1.0.8 |
Solaris:AIX |
Replace |
mqseries/end_mq |
|
4.1.0.8 |
Solaris:AIX |
Replace |
mqseries/start_mq |
If you have Collbutils.jar from
version 4.0.1 or 4.1.0.x, you need to pick the latest Collabutils.jar file from
4.1.0 fixpack for the RosettaNet collaborations to work .
© 2002 IBM Corporation. Proprietary and Confidential. All Rights
Reserved.