When attempting to start WebSphere MQ Workflow after installation with Oracle V10.2 the Workflow Admin Server(FMCAMAIN) dumps a core and the Workflow system fails to start.
WebSphere MQ Series is seen to start fine. When FMCAMAIN is executed it ends immediately with a core dump. The log for FMCZINSX, the initial configuration utility, shows unsuccessful completion when running 'fmczinsx -o oracle10'.
WebSphere MQ Workflow is linked to the wrong Oracle database libraries.
Diagnosing the problem
Examine the log fmczinsx.log after running FMCZINSX to set up the environment for Oracle. An unsuccessful completion of 'fmczinsx -o oracle10' may look like the following:
FMC00100 17:20:25 Values from installation profile /opt/fmc/fmcrc successfully read.
FMC00150 17:20:25 System wide Workflow setup changed to support database Oracle 10g
FMC00155 17:20:25 Checking availability of required shared libraries for database Oracle 10g
FMC00157 17:20:25 Using library search path /opt/fmc/lib/workflow
FMC00157 17:20:25 Using library search path /usr/lib
FMC00157 17:20:25 Using library search path /opt/fmc/lib
FMC00157 17:20:25 Using library search path /opt/oracle/product/10.1.0/lib32
FMC00158 17:20:25 --> Shared library libclntsh.sl.10.1 could not be loaded using the builtin search paths!
FMC00159 17:20:25 Please ensure that libclntsh.sl.10.1 is available if this database setup will be used.
FMC00170 17:20:25 +++ WARNING +++ Function completed with warnings +++
Resolving the problem
Check the links in <WorkflowInstallDirectory>/fmc/lib/workflow for libclntsh.sl and libclntsh.sl.10.1. Make sure they are pointing to instances of shard libraries in $ORACLE_HOME/lib32 and not $ORACLE_HOME/lib. After correcting run 'fmczinsx -o oracle10' again and check results in fmczinsx.log.
WMQ MQ WF