Creating SOA on OCI From Marketplace Image

User Personalization With MDS in Oracle ADF

(ADF/JDeveloper/Weblogic 12.2.1.3, Oracle Database 12.1.0.2)

In this post, I want to list the steps involved in setting up the User Personalization Configuration for an ADF Application with MDS Repository in the Database.

This is all documented very well in Oracle Documentation and a lot of blog posts, but most of the information is scattered across different sources. The errors which are encountered while implementing this may not give much information to find out the actual issue and it can be sometimes frustrating. The error is mentioned in point 7 below.

In this post I will try to put all of the steps together so that anyone who is trying to do this for the first time will find it easy to implement it. I personally found it difficult to find what I was missing and it took a lot of research and help form others to get this working. This particular gap I am mentioning below in the steps 4 and 5

Assuming that you have already done the following ***

ADF Application is enabled with Security

Deployed on a clustered Weblogic Server environment

MDS Repository is already created

Data Source is configured in Weblogic Console.

Personalization will work only if security is enabled as the personalizations are stored against each authenticated user.

1.      First we need to enable the User Customization Settings in the View Controller Project Properties

2.      Then we need to enable the MDS Customization Settings for the ADF Application

This should be done in adf-config.xml file

We may have to go to the source of the file to change the XML as below for the section ‘adf-mds-config’

  <adf-mds-config xmlns="http://xmlns.oracle.com/adf/mds/config">

    <mds-config xmlns="http://xmlns.oracle.com/mds/config" version="11.1.1.000">

      <cust-config>

        <match path="/">

          <customization-class name="oracle.adf.share.config.UserCC"/>

        </match>

      </cust-config>

      <persistence-config>

        <metadata-namespaces>

          <namespace path="/persdef" metadata-store-usage="MAR_TargetRepos"/>

        </metadata-namespaces>

        <metadata-store-usages>

          <metadata-store-usage id="MAR_TargetRepos" deploy-target="true" default-cust-store="true">

            <metadata-store class-name="oracle.mds.persistence.stores.db.DBMetadataStore"> 

              <property name="jndi-datasource" value="jdbc/mds/DEV_MDS"/> 

              <property name="repository-name" value="mds-APP_MDS"/>

              <property name="partition-name" value="APP_PARTITION"/>

            </metadata-store> 

          </metadata-store-usage>

        </metadata-store-usages>

      </persistence-config>

    </mds-config>

</adf-mds-config>

3.      Select the ADF Faces Components you want the User Personalization to be enabled for

At this stage when we run the application in JDeveloper with Integrated WLS, it may work well as it might use the File based configuration for storing the user personalization settings.

But if you are using an Ant Build Script or Deploy the application from WLS console to a clustered server, then you will have to do one more additional step. And this was what I was missing and took me a while to figure.

4.      We need to create and register a Metadata Repository along with a Partition for the Application

This can be done either manually at the command prompt or using a ‘python’ script.

For this we need to use the Weblogic Scripting Utility that comes along with the JDeveloper installation.

C:\app\oracle\Middleware\jdev12213\oracle_common\common\bin\wlst.cmd

When we run this utility it will give us a set of APIs to call to create and register the repository.

The general steps involved are as follows

                    Connect

                    Register Metadata Repository

                    Create Partition

For Example

    connect(<WLS Admin User>, <WLS Admin Pwd>,  <WLS Admin Console URL>)

    registerMetadataDBRepository(name='mds-APP_MDS', dbVendor='ORACLE', host=<WLS Host>, port=<WLS Port>, 

    dbName=<DB Service Name>, user=<DB Username>, password=<DB Password>, targetServers=<Cluster Name>);

   

    createMetadataPartition('mds-APP_MDS', 'APP_PARTITION');

   

    exit()

   

This can be run on any machine as long as we give the correct server parameters

This is typically a one time operation.

5.      Update the Application EAR file with the Metadata Repository information

Once the application EAR file is built, we need to update the EAR file with the MDS Repository and Partition Information. For this also, we need to use the WLST commands or write a python script.

The general steps involved are as follows

                    Connect

                    Open EAR file

                    Set Application Metadata Repository Information

                    Save the EAR file

For Example

    connect(<WLS Admin User>, <WLS Admin Pwd>,  <WLS Admin Console URL>)

   

    archive = getMDSArchiveConfig(fromLocation=<EAR File Path>)

            

    archive.setAppMetadataRepository(repository='mds-APP_MDS', partition='APP_PARTITION', type='DB', 

    jndi='jdbc/mds/APP_MDS')

   

    archive.save()

   

6.      Now, the EAR file is ready to be deployed to a clustered server with User Personalizations enabled with MDS Repository

7.      Error when deploying without updating the EAR file

     [exec] weblogic.deploy.api.tools.deployer.DeployerException: Task 0 failed: [Deployer:149026]deploy application APP on DefaultServer.

     [exec] Target state: deploy failed on Server DefaultServer

     [exec] weblogic.application.ModuleException: oracle.mds.config.MDSConfigurationException: MDS-01335: namespace "/persdef" mapped to metadata-store-usage "MAR_TargetRepos" but its definition was not found in MDS configuration. :oracle.mds.config.MDSConfigurationException:MDS-01335: namespace "/persdef" mapped to metadata-store-usage "MAR_TargetRepos" but its definition was not found in MDS configuration.

     [exec]     at oracle.mds.config.PConfig.populateNamespaceConfigList(PConfig.java:1499)

     [exec]     at oracle.mds.config.PConfig.loadFromBean(PConfig.java:1166)

     [exec]     at oracle.mds.config.PConfig.<init>(PConfig.java:892)

     [exec]     at oracle.mds.config.MDSConfig.loadFromBean(MDSConfig.java:1288)

     [exec]     at oracle.mds.config.MDSConfig.loadFromElement(MDSConfig.java:1360)

     [exec]     at oracle.mds.config.MDSConfig.<init>(MDSConfig.java:905)

……

8.      Reason for the above error is that even though we have added the MDS Configuration in the adf-config.xml file, when the EAR is built initially, this entry will be removed by Ant. This is the expected behavior according to oracle documentation.

So we are supposed to perform the additional step I mentioned in the step 5 to update the EAR file with the repository information.

9.      Some Observations

a.      User Personalization Settings will persist across multiple deployments of the application

b.      The EAR file size that is updated by the WLST API in the step 5 is much lesser than the original EAR

c.      Once the User starts personalizing the pages, after that if we make changes to the adf-config.xml to remove some components from personalization, the change does not seem to be effective.

Sub Query Workarounds for ADF View Object Design Time Issues in JDeveloper

ADF/JDeveloper 12.2.1.3

In this blog, I will show some workarounds for some specific issues we may see in JDeveloper View Object Wizard

When we are designing a View Object in JDeveloper, especially when we modify the query in the expert mode, in some cases as mentioned below, the wizard does not build the attribute mappings as expected.

• When the Query has an 'extract' clause
• When the Query has a table or view accessed from a DBLink 
• When the Query has a synonym for an object from another Schema

We normally see this behavior when we join attributes from tables with one of the above. 

A General Workaround for these kind of issues is to use a Sub Query. 

• When we use Extract clause in the query to get an attribute, the attribute is not shown as a selected attribute in the Attribute Mappings. This issue comes especially when we join the ‘extract’ attribute with other attributes

• When we query a table via a DB Link the attributes from the DB Link are not shown in the ‘Attribute Mappings’. This issue comes especially when we join the ‘dblink’ attributes with other attributes

• When we query columns from a partitioned table the attributes are not shown in the Attribute Mappings. Below table is partitioned on ‘payroll_date’.This happens especially when we join the attributes with other table attributes.Other best option in this case is if the table is in a different Schema, then we may have to give the Owner Schema Prefix also

Shared ADF Libraries Within One Application

(JDeveloper 12.1.3, WLS 12.1.3)

This post is inspired from the original post by Andrejus Deploying ADF Applications as Shared Libraries on WLS.

I tried to do the same thing but the only difference is that I tried to keep all the code in one single ADF application.

I think this will keep all code related to the whole application in one place in source control, so maintenance will be easy.

This will also eliminate the necessity of the SharedLib intermediate Application.

And we can deploy the shared libraries changes individually without affecting other parts of the application.

And ofcourse these shared libraries can be accessed from a different ADF application as well.

Departments and Jobs ViewController projects are deployed as Shared ADF Libraries.

HRDashboard project is the main project which will use the two libraries.

Create 2 Deployment profiles in Departments and Jobs ViewController projects as shown below, one an ADF Library Jar profile and another WAR profile.

In the HRDashboard Project, create a weblogic.xml deployment descriptor to add the 2 shared libraries we created above in the Departments and Jobs projects. The names should be same as the WAR profile names.

Add both the library jar files to the HRDashboard project.

Create a WAR deployment profile for the HRDashboard project.

Remove the Contributors for Departments and Jobs in the Library filters of the WAR deployment profile.

Create an EAR Application Deployment Profile by selecting only the HRDashboard WAR deployment profile in Application Assembly.

Deployment Sequence

1.      Deploy both the Departments and Jobs ADF Library  profiles to build the JAR files.

2.      Deploy both the Departments and Jobs WAR profiles to the Weblogic server as Shared Libraries.

3.   Deploy the EAR profile to the Weblogic Server as a standalone application.

Note : It looks like we have to Stop the main Application in the Weblogic Console when we are redeploying any of the shared libraries, otherwise it will not allow us to deploy the shared libraries. We can start it as usual when the shared library deployment is completed.

Human Task Assignees

(JDeveloper 12.1.3, WLS 12.1.3, SOA/BPM Suite 12.1.3)

In this blog, I will show some of many possible ways to assign a Human Task to one or more Assignees and observe their behavior.

If a task has to be assigned to more than one user, the simplest way is to assign it to a Group, but only if all the users are in that Group. But if there are some users in the group which the task should not be assigned to, or not all the users are present in the group, then assigning the task to a Group may not work exactly how we want it to.

In such cases we can set the assignees to a task in any of the following ways.

1.      Separate Parallel Participants within same Stage

We can modify the Human Task Assignees to add a new Participant as shown below. In this example, I will set the first participant to be derived from the Swimlane. (User is assigned in Organization file)

And the second participant to be derived from the payload.

Both the above assignees are defined in the same stage, so considered to be Parallel. That means the task will be assigned to both the assignees at the same time.

When one of the assignees responds to the task, the task is still assigned to the second user and is not automatically completed. 

The second user also has to complete the task, only then the whole human task is set to completed status. After the second user completes the task, below is the human task status.

Note that the ‘Voting’ concept is not applicable here since the task assignee Type is still ‘Single’. So the human task is not completed until all the assignees complete their tasks. (Voting is applicable when the Assignee Type is ‘Parallel’)

Also, Note that ‘Claim’ also will not be applicable here. Even if the first user Claims the task, the second user also have to submit the task. Until then, the human task is not set to completed status. (Claim is applicable when assigned to a Group or a comma separated users)

2.      Separate Serial Participants within same Stage

For this I will just change the participants sequence as shown below.

In this case when the task is created, the task will be first assigned to the Participant1.

Only when first participant submits, the task will be then assigned to the Participant2.

Only when all the assignees in the chain submit their tasks, the human task is set to completed status.

3.      Task Assignee Participant Type : Parallel

For this I will change the Participant type to ‘Parallel’ and get both the assignees from the Payload.

In this, the task will be assigned to all the assignees at the same time.

For Parallel type of assignee, we can specify the task completion condition by vote percentage in the Voting tab.

Following condition means that if 50% of the assignees vote, irrespective of the outcome, the human task is said to be completed.

For example, if we set the following condition, and both the assignees select REJECT action, then the default outcome SUBMIT will be applied to the human task.

Note that the ‘Claim’ option is not available for the assignee in this case.

4.      Task Assignee Participant Type : Serial

Works similar to 2. (Separate Serial Participants within same Stage)

5.      Comma Separated Users

For this I will set the participant from only one payload attribute, but pass a comma separated assignee name value to it while calling the process.

In this case also, the task is assigned to both the users at the same time.

Note that the Claim option is available to the users in this case.

When one user ‘Claims’ the task, the other user will only be able to see the task in his queue but will not be able to perform any actions on the task.

When any one of the users submits the task, the human task is completed and is removed from the other users Queue.

It is logically similar to the behavior when both the users are in a Group and the task is assigned to the Group.

6.      Group Assignees

For this I created a group ‘sdarbhaGroup’ in WLS and added 2 users sdarbha1 and sdarbha2 in the group.

The behavior is similar to 5. (Comma Separated Users)

7.      Complex Assignees

This is a very powerful way to set assignees based on complex business requirements.

Let us assume that a particular human task has to be performed by many users with different business responsibilities and in a particular sequence etc.

For example, a task has to be completed by a data entry operator who enters some basic information, then has to be assigned to 2 of his peers and a group for review, and further has to be sent over a management chain, and finally a notification to the administrator.

First assign it to a data entry operator.

Then assign it to his 2 peers.

Along with a group at the same time.

Then send it to a management chain for review.

And finally send a notification to the Administrator.

When the task is generated following is the sequence of assignees.

The Final FYI Notification will not be visible in the task flow audit trail, but is sent to the administrator and will be visible in the BPM Workspace.

8.    Multiple Participants of different types in a Single Task Assignee Type

We can also assign the task to multiple users, groups and application roles, 

keeping the Assignee type as Single.

In this case the task will be assigned to all the users, users in the group,
and users from the application at the same time.
Any one from the above users can claim the task and work on it.
The task will be marked as completed when one user submits the task.

9.    Multiple Participants of different types in a Serial Task Assignee Type

We can also assign the task to multiple users, groups and application roles, 

keeping the Assignee type as Serial.

In this case first the task will be assigned to sdarbha3 only. 

When he submits, then only the task will be assigned to sdarbha4.

Even though the users sdarbha3, sdarbha4 are assigned at the same time in the task definition, when the task is created, they will be assigned one after the other since the task type is Serial.

But, the task is assigned to the group, all the users in the task will be assigned at the same time and anyone can claim the task and work on it.

Creating and Executing Test Cases for BPM Processes

(JDeveloper 12.1.3 / WLS 12.1.3 / SOA/BPM Suite 12.1.3)

  

In this post I will show how to create and execute Test Cases for BPM Processes.

I will be using the same project which I used to show how to define Business Rules in BPM.

http://sameerdarbha.blogspot.com/2013/07/implementing-business-rules-in-bpm.html

The above project was created in 11.1.1.7 version of BPM.

For this post, I will be using JDeveloper/BPM Suite/WLS 12.1.3.

I have migrated the application ‘EmployeeTravelExpenseSystem’ to 12.1.3.

And I modified the process a little bit to add another Start activity of type ‘Message’ so that we can start and test this process from EM console. It takes the input as the same object ‘TravelExpenseObject’ as shown below.

Right click on ‘SOA > testsuites’ to create a Test Suite as shown below and give a name to the Test Suite.

Right click on the Test Suite ‘ETEAPTestSuite’ and create a Test case ‘AutoApprovalTestCase’

Click ‘Generate Sample’ button to generate the input XML and enter the parameters satisfying the Auto approval case according to the Business rules. (Check the previous post to check the conditions).

A new file ‘AutoApprovalTestCase.xml’ file will be created as shown below which looks like the composite.xml

Create another test ‘RequiredApprovalTestCase’ with the input xml as follows satisfying the Required  Approval case according to the Business rules.

We can run the individual Test cases or even the complete Test Suite from within JDeveloper.

Right click the Test Suite name and select ‘Run Test Suite’.

JDeveloper will ask for the server name to which this application has to be deployed. Enter the name of the server.

 Enter the test case execution name and select the test cases you want to run.

 A new file will be opened showing the test results of the Test Case execution.

Also these test case execution results can be seen in the EM console.
This application is deployed to a 'Compact Domain' in the WLS.

Optionally the test cases also can be run on the EM console, since the test cases are deployed along with the application.

When the Test Suite is run, 1 instance of the process for each test case is created.
The instances created by the Automated Test Cases like this will be shown with a yellow 'dot'.

The Test case for Auto approval process flow is as follows.

The test case for Approval required process flow is as follows.