Oracle® Fusion Middleware Release Notes 11g Release 1 (11.1.1) for IBM AIX on POWER Systems (64-Bit) E14771-44 |
|
![]() Previous |
![]() Next |
This chapter describes issues associated with Oracle Data Integrator. It includes the following topics:
Section 40.2, "Design-Time Environment Issues and Workarounds"
Section 40.3, "Oracle Data Integrator Console Issues and Workarounds"
Section 40.4, "Technologies and Knowledge Modules Issues and Workarounds"
Section 40.5, "Oracle Data Profiling and Oracle Data Quality Issues and Workarounds"
This section describes the following issues and their workarounds:
Section 40.1.1, "Generated Templates Including a Datasource With an Oracle RAC URL Fail to Deploy"
Section 40.1.2, "Generated Templates Including Datasources Fail to Deploy"
Oracle WebLogic Configuration Wizard does not support Oracle RAC URLs for datasources. When a datasource is included in a Java EE Agent template that has been generated using Oracle Data Integrator Studio, this template fails to deploy, and raises the following exception is raised:
com.oracle.cie.wizard.WizardController - Uncaught Exception java.lang.NullPointerException atcom.oracle.cie.domain.jdbc.DatasourceXBeanAspectHelper.decomposeURL
As a workaround, do not include this datasource definition in the generated template and deploy the datasource from Oracle Data Integrator Studio.
As the Oracle WebLogic Configuration Wizard uses specific code for parsing the URL and exposing its parameters in a user friendly format, it only supports a predefined set of JDBC URLs. If a JavaEE Agent template that has been generated using Oracle Data Integrator Studio contains a datasource with a URL or driver not included in this set, the Wizard fails to use this template with the following exception:
com.oracle.cie.wizard.WizardController - Uncaught Exception java.lang.NullPointerException atcom.oracle.cie.domain.jdbc.DatasourceXBeanAspectHelper.decomposeURL
As a workaround, do not include this datasource definition in the generated template, and deploy the datasource from Oracle Data Integrator Studio.
The Clean Stale Sessions feature of the Agent relies on the stale or dead connections of the connection pool to be automatically cleaned up either through expiration or connection validations tests done by the connection pool implementation.
If the stale connections are not removed out by the connection pool, then the Clean Stale Sessions feature does not work even after the database has been restarted. None of the other requests of the agent will work after the database restart if the agent itself is not restarted.
In a WebLogic Server container, connections are validated every few seconds using a test query. It is recommended to verify and provide the table name used by the database to run the test query: select count(*) from <test_table_name>
.
For Master Repositories and Work Repositories, datasources use SNP_LOC_REP
and SNP_LOC_REPW
as the test tables.
Alternatively, you can use the DUAL
table for an Oracle repository.
This section describes the following issues and their workarounds:
Section 40.2.1, "For File Datastores, the View Data operation always looks for global context"
Section 40.2.2, "Quick Edit Editor deletes incorrect Joins when sorted"
Section 40.2.4, "Inconsistent Behavior when Refreshing Load Plans"
Section 40.2.5, "User Parameters are not Taken into Account by the Runtime Agents"
Section 40.2.6, "Incorrect Non-ASCII File Names in Zip Export Files"
Section 40.2.7, "Non-ASCII Characters Incorrectly Converted to Underscore Characters"
Within ODI Studio, when viewing data from within a data store, there must be a global context defined for the datastore.
com.sunopsis.core.SnpsInexistantSchemaException: ODI-17523: There is no connection for this logical schema / context pair: lg1 / GLOBAL.
As a workaround, define a global context for the datastore.
Within ODI Studio Quick Edit, when deleting a join, the incorrect join may be deleted if a join is selected and then the list of joins is sorted.As a workaround sort the join list prior to deleting the join with Quick Edit.
Within ODI Studio Quick Edit, when deleting a source, the source will appear as removed even if "No" is selected from the warning before the deletion.
As a workaround, refresh ODI Studio and open Quick Edit, the source will still be available.
Within ODI Studio, when refreshing Load Plan to pick latest Scenario reference in the Load Plan, the latest scenario is determined by the Scenario Creation timestamp. During ODI agent execution, the latest scenario is determined by alphabetical ascending order of the Scenario Version string value.
The runtime agents do not take into account the user parameter values specified in the userpref.xml
file. The default values are always used instead.
There is no generic workaround at this time.
Non-ASCII file names are incorrectly generated in Zip export files. Non-ASCII characters in these file names are replaced with underscores. Setting the Never transform non ASCII characters to underscores user parameter to Yes
has no effect. As a workaround, export the files in a folder and compress this folder.
Non-ASCII characters are handled incorrectly in the following situations, even if the Never transform non ASCII characters to underscores user parameter is set to Yes:
Non-ASCII characters are replaced with underscores in the Alias field of a column when using the fixed file reverse-engineering wizard.
Non-ASCII characters from the header line are replaced with underscores in column names when doing a delimited file reverse-engineering.
As a workaround, manually modify the column definitions after the reverse process.
This section describes the following issues and their workarounds:
Section 40.3.1, "Oracle Data Integrator Console Page Task Flow Resets When Another Tab Is Closed."
Section 40.3.2, "Enterprise Manager Logout Does Not Propagate to Oracle Data Integrator Console"
When the tab containing a page is closed in Oracle Data Integrator Console, all the tabs after this one reset their task flows (the navigation actions that were made). The pages in these tabs reset to the original object that was opened.
There is no workaround for this issue at this time.
When Enterprise Manager Fusion Middleware Control (EM) is used in conjunction with Oracle Data Integrator Console with Single Sign-On (SSO) activated, a user logging out from SSO in Enterprise Manager might still be able to access Oracle Data Integrator Console pages from the same browser window by typing a direct URL.
This issue is caused by a cookie that is created by Oracle Data Integrator Console not being invalidated by the Enterprise Manager logout action.
As a workaround:
Administrators should configure their Enterprise Manager and Oracle Data Integrator Console servers for access fronted by a gateway. The gateway automatically redirects the user that is not logged into SSO to the SSO Login page.
If you're accessing Oracle Data Integrator Console or Enterprise Manager from a shared or public computer, close the session by logging out. For additional security, close the browser session.
This section describes the following issues and their workarounds:
Section 40.4.1, "For Microsoft Excel Target Datastores, column names must be in upper case"
Section 40.4.2, "For File Datastores, reverse-engineering process cannot be canceled"
Section 40.4.3, "SQL Keywords are not imported when importing a Technology in Duplication Mode"
Section 40.4.6, "IKM and LKM SQL Incremental Update (Row by Row) fail with Non-ASCII Characters"
Section 40.4.7, "IKM SQL Incremental Update (Row by Row) Cannot Target Japanese Tables"
When using Microsoft Excel as a target, column names must be in upper case.
[Microsoft][ODBC Excel Driver]COUNT field incorrect
As a workaround, rename column names within Microsoft Excel.
Within ODI Studio, when reverse engineering new columns to an existing datastore, after reverse new columns are automatically saved when answering "No" to "<datastore> has been modified. Save changes?"
Within ODI Studio, when importing a technology in duplication mode, the SQL Keywords are not imported.As as workaround, manually add the SQL Keywords to the technology properties.
When running interfaces and using a XML or Complex File schema as the staging area, the "Unknown Token" error appears. This error is caused by the updated HSQL version (2.0). This new version of HSQL requires that table names containing a dollar sign ($) are surrounded by quotes. Temporary tables (Loading, Integration, and so forth) that are created by the Knowledge Modules do not meet this requirement on Complex Files and HSQL technologies.
As a workaround, edit the Physical Schema definitions to remove the dollar sign ($) from all the Work Tables Prefixes. Existing scenarios must be regenerated with these new settings.
When the Model code contains non-ASCII characters, adding subscribers does not work for Consistent Set Journalizing KMs. This applies to all Consistent Set JKMs. The Session completes successfully but no subscriber is added. Note that this issue does not occur when the Model code contains exclusively ascii characters.
There is no generic workaround at this time.
When the source datastore contains non-ascii characters the interface using the IKM or LKM SQL Incremental Update (Row by Row) fails at the step Insert flow into I$ table with the following error message:
org.apache.bsf.BSFException: exception from Jython: Traceback (most recent call last): File "<string>", line 149, in <module> UnicodeEncodeError: 'ascii' codec can't encode characters in position 6-7: ordinal not in range(128)
There is no generic workaround at this time.
IKM SQL Incremental Update (Row by Row) fails on task Flag rows for update if the target table name contains Japanese characters. In the generated code, the integration table name appears as "I$_??????" where "?" characters replaces a Japanese character.
As a workaround, replace this IKM with another IKM suitable for the technology. For example, 'IKM Oracle Incremental Update' for Oracle.
This section describes the following issues and their workarounds:
Section 40.5.1, "ODIDQ fails to Create Loader Connection when Server is Installed on Linux 64 bits"
Section 40.5.2, "Unable to Connect the Client to a Data Quality Server on UNIX"
Section 40.5.3, "Menus Are in English When Starting the Client Using a .tss File."
When trying to create loader connections with ODQ installed on Linux 64 bits server, ODQ Client installed on Windows machine, and the Metabase Host set as the Linux Server, the loader connection is not created and no error message or warning is displayed.
As a workaround, download the patch 11775170 and follow the instructions in the README file included with the patch:
Go to My Oracle Support:
Click on the Patches & Updates tab.
In the Patch Search area, search for patch 11775170.
Download the patch.
When connecting from the Oracle Data Quality Client to a Data Quality Server installed on a UNIX machine the following error message may appear: "No support for client application "Admin" version 12.0.1".
To solve this issue re-install the metabase definitions using the following steps:
On the UNIX server go to the following directory:ODQ_HOME/oracledi/odp/Client/SvrSide
Run ./setup
At the prompt, enter the Metabase Installation location: ODQ_HOME/oracledq/metabase_server/
Enter the metabase admin user and password.
When the metabase definitions installation completes, restart the Scheduler service using the following command: ODQ_HOME/oracledq/metabase_server/metabase/bin/scheduler -restart
When starting Oracle Data Quality Client by double-clicking on a saved .tss file, the menus appear in English independently of the client machine's locale.
As a workaround, open the client from the shortcut menu, then open the .tss file.