SOA Suite 10.1.3.4 loses instances in the BPEL console

Question

Hello,
One of my BPEL processes loses instnaces in the BPEL console.
OS: AIX 5.3 oslevel –s:5300-04-02
SOA Suite 10.1.3.4
I feel I hit the same problem as this thread, process is supposed to be in a jta transaction
But I have set following 3 values (all in seconds), and I still have the problem. The SYNC call is to query the Oracle EBS PO by PO NO, and get its po lines, line shipments, and shipment distributions by Oracle Database Adapter. If my PO just had one line, the process instance can be seen in the console, otherwise it won't, but all these instances are finished, because in the end the process will enqueue a message into a queue. And I can see all these messages there. Anyone can help? Thanks!
$ORACLE_HOME/j2ee/oc4j_soa/config/transaction-manager.xml
transaction-timeout="240"
$ORACLE_HOME/j2ee/oc4j_soa/application-deployments/orabpel/ejb_ob_engine/orion-ejb-jar.xml
transaction-timeout="180"
$ORACLE_HOME/bpel/domains/default/config/domain.xml
<property id="syncMaxWaitTime">
<name>Delivery result receiver maximum wait time</name>
<value>600</value>
-------------- Error messages ----------------
<2009-04-02 11:40:03,319> <ERROR> <default.collaxa.cube.engine> The execution of this instance "40002" for process "PROEBSMsgToB2B" is supposed to be in a jta transaction, but the transaction is not present or in active state, please turn on the application server transaction debug logs to get more information.
<2009-04-02 11:40:03,324> <ERROR> <default.collaxa.cube> <BaseCubeSessionBean::logError> Error while invoking bean "cube delivery": Scope not found.
The scope "BpPrc0.1" has not been defined in the current instance.
ORABPEL-02094
Scope not found.
The scope "BpPrc0.1" has not been defined in the current instance.
Regards,
Kerr
Edited by: Kerrs on 2/04/2009 12:15

Answer 1

Hi Kerr,
These are the possible cause to the problem and their solutions:
Poor performance of the dehydration database If you are using Oracle Lite as dehydration store, please switch to use Oracle 9i or 10g. If Oracle 9i/10g is already in use, check the database parameter 'process' and 'session' to make sure it can handle the expected throughput.
OC4J has too few available connections to the dehydration database. Increase the maxConnection number of the BPELServerDataSource at the BPEL_HOME/integration/orabpel/system/appserver/oc4j/j2ee/home/config/data-sources.xml (for developer edition) or IAS_HOME/j2ee/OC4J_BPEL/config/data-sources.xml (mid-tier installation).
Size of message is too big Two ways to deal with this problem:
Increase the transaction timeout at PEL_HOME/integration/orabpel/system/appserver/oc4j/j2ee/home/config/server.xml (developer edition) or IAS_HOME/j2ee/OC4J_BPEL/config/server.xml (mid-tier installation) : http://download-west.oracle.com/docs/cd/B31017_01/integrate.1013/b28981/app_trblshoot.htm#sthref3957
Decrease the auditLevel from BPELConsole -> Manage BPEL Domain -> Configurations tab. Doing so will reduce the amount of data saved to the dehydration store.
Having said all that , its always good to be on latest 10.1.3.4 MLR # 5, so that if you have an SR logged with Oracle Support, they will not again ask you to upgrade to latest patch, rather they can focus on getting a fix for you on top of the latest patch itself.
Hope this helps!
Cheers
Anirudh Pucha