This topic applies only to the IBM Business Automation Workflow Advanced
configuration.

Adding support for shared work items

If you migrated your Business Process Choreographer configuration from Version 7.0.0.2 or earlier, you can improve the database performance by activating shared work items. Any new Business Process Choreographer configurations automatically support shared work items.

About this task

Depending on how many work items you currently have in your Business Process Choreographer database, and how many hours per day you can run the work item migration script, this task might take several days to complete.

Procedure

Optional: Reduce the number of work items in your system by completing one or more of the following steps:
- Deleting BPEL process instances
- Deleting completed task instances
- If you configured a Business Process Archive Manager to archive instances from this Business Process Choreographer configuration, complete the actions that are described in Archiving completed BPEL process and task instances.
At a time when there is a low load on the database, run the migrateWI.py script to create shared work items for existing work items.
1. Decide on the maximum duration in minutes, duration, that you want the script to run for.
2. Change to the directory where the migrateWI.py script is located:
  - install_root/ProcessChoreographer/admin
  - install_root\ProcessChoreographer\admin
3. Run the script by starting wsadmin with the following parameters:
```
  -conntype NONE
 [-profileName profile] 
 [-tracefile trace_file] 
  -f migrateWI.py 
  -migrateToWISharing
 [-duration duration]
 [-slice slice_size]
 -cluster clusterName
[[-dbUser userID] -dbPassword password]
 [-dbSchema schema]
```
  Note: You must run the script on the node of a cluster member, and not on the deployment manager. Because wsadmin overwrites its trace file, use the -tracefile option to specify a file name and location for the trace file for the work item data migration. You do not need to stop the cluster. For more information about using this script and its parameters, see Business Process Choreographer data migration script for shared work items.
4. When the script finishes, it reports a summary of the migration status of the work items that belong to each type of entity:
  - The total number in the system.
  - The number that were migrated.
  - The number that still must be migrated.
  - The percentage that are completed.
  For example:
```
                   Entities  Migrated  Not migrated  Completed
EventInstance          7602      7602             0       100%
WorkBasket                0         0             0       100%
TaskInstance          29883     29883             0       100%
EscalationInstance     4227      4227             0       100%
BusinessCategory          0         0             0       100%
ActivityInstance      52572     52572             0       100%
ProcessInstance        7602      3036          4566        39%
--------------------------------------------------------------
Total                101886     97320          4566        95%
```
If the script finished with less than 99% completed (that is, more than 1% of the total number of entities and their corresponding work items in the system still need to be processed), repeat step 2.
If the script finished with 99% or more completed, activate the shared work item support.
1. Stop the cluster where Business Process Choreographer is configured.
2. Now run the script a final time to process any remaining work items and switch into shared work item mode, by starting wsadmin with the following parameters:
```
  -conntype NONE
 [-profileName profile] 
 [-tracefile trace_file] 
  -f migrateWI.py 
  -setMode WS_SHARING_ACTIVE_MODE
  -cluster clusterName
[[-dbUser userID] -dbPassword password]
 [-dbSchema schema]
```
3. Success is indicated by the following message:
```
migrateWI.py finished with warnings.
```
  In case of problems, you might get a message similar to the following one:
```
Exception received:
java.lang.IllegalStateException: Mode: WS_SHARING_ACTIVE_MODE
migrateWI.py finished, result = WARNING. 
```
4. Restart the cluster.

Results

The Business Process Choreographer configuration is using shared work items and has better performance for many database queries.

What to do next

Optionally, you can perform any of the following steps:

Use the administrative console to change the default cleanup times and retention periods for unused shared work items. You can change these settings on the Runtime or Configuration tab for the Human Task Manager.
Delete unused staff query interfaces and unused work item patterns, by running the cleanupUnusedStaffQueryInstances.py script.
If you do not use any queries that must run against the non-shared work items, you can free up space in the database by customizing and running a copy of the dropClassicWIIndexes.sql script to drop the indexes on the WORK_ITEM_T table.
1. The script is in the following location.
  - install_root/BPM/dbscripts/database_type/Tuning
  - install_root/BPM/dbscripts/database_type/Tuning
2. Make a copy of the script file dropClassicWIIndexes.sql.
3. Edit your copy.
  1. Replace all occurrences of the string @SCHEMA@ with the name of your schema. If you use an implicit schema, you must delete the string @SCHEMA@., including the trailing dot.
  2. If you are using a DB2® for z/OS® database, replace all occurrences of the string @STOGRP@ with the name of your storage group.
  3. If you are using an Oracle database, replace all occurrences of the string @INDEXTS@ with the name of the index table space.
  4. Save your changes.
4. Run your edited copy of the script. In your database client command line processor, enter one of the following commands:
  - For DB2: db2 -tf dropClassicWIIndexes.sql
  - For Oracle: sqlplus dbUserID/dbPassword@database_name @dropClassicWIIndexes.sql
  - For Microsoft SQL Server: sqlcmd -U dbUserID -P dbPassword -e -i dropClassicWIIndexes.sql
  For more information about how to run scripts on your database, see the product documentation for your database. If there are any errors, or failure is indicated in your database client output, fix the reported errors and repeat this step.