Published Release Notes

EAR support for JBoss EAP 6

PRPC deployment in JBoss EAP 6 as an EAR archive is now supported. 

If you need to deploy the JBoss EAR file, go to My Support Portal and submit a Support Request. GCS can assist you with the procedure.

Small distribution kit

A new “small” release distribution format for Maintenance Levels (ML) is now available, designed to significantly increase adoption via: 

• Minimal package size and faster download time
• Minimal install time via tactical changes to the upgrade implementation and process

This kit contains streamlined installers (both UI and command-line installers) and specialized error handling for ML updates (for example, a “small” kit will only work if an existing version of the same release is already installed). 

The kit assembly has also been automated, which includes the construction of a PRPC_Rules.jar archive containing cumulative rules since the GA release. 

A “small” kit will work for any prior ML release within a major release (for example, the “small” ML5 distribution kit will update any customer on 7.1 ML1 through 7.1 ML4).

Improved data page performance

Improvements have been made with queueing and processing of asynchronously-loaded data pages to reduce the number of calls to the database and avoid generating performance problems because of unneeded asynchronous requests.

Restricted SMA access

System Management Application (SMA) remote connections are not supported on JBoss EAP 6.x. This means SMA users may only monitor locally connected systems. Ensure that the "Local JVM Connection..." option is selected when configuring PRPC nodes:

DB2-LUW database logfile size increase

To avoid running out of logfile space due to large transaction sets during the rule base load of a Pega 7.1.x install, upgrade, or maintenance level update, systems supported by a DB2-LUW database platform should increase the LOGFILSIZ parameter to at least 4096 pages from the default size of 1000 pages.

After the size has been increased, restart the database to ensure that the new setting is loaded into the database correctly.

Need to run script before updating Multitenant systems

When updating or upgrading a Multitenant system from Pega 7.1.5 or 7.1.6 to Pega 7.1.7, if that system uses either an Oracle or a PostgreSQL database, you may encounter the error:

“Table must be empty to add column.”

The Multitenant architecture requires an additional column on a number of the PRPC database tables (“pzTenantID”). In Pega 7.1.7, two additional PRPC tables were tenant-qualified: pc_schedule_task and pr_index_schedule_task. The Multitenant column is added to these tables by the update/upgrade process. However, Oracle and PostgreSQL do not allow the addition of a non-null column to an existing table unless the table is empty, so updating or upgrading systems on those databases displays the error detailed above.

To avoid this error, before beginning the update or upgrade, it is necessary to run a script:

• Oracle — OracleMTupgrade.sql
• PostgreSQL — PostgresMTupgrade.sql

For updates, these scripts are located in the /scripts/ddl directory.

For upgrades, these scripts are located in the /Resourcekit/AdditionalUpgradeScripts directory.

Specify the scope for rolling back rules and data to a restore point

Create restore points to save the state of all rules and data in your system at a significant point in time, for example, before you import an application. Roll back to that restore point to return the system to that state. Now, you can filter which rule and data instances are returned to their previous state:

• System: Roll back every rule and data instance that has a history record.
• User: Roll back rule and data instances modified by a specific user. If any rule was changed by more than one user, you will see an error message and must use the system rollback.
• Application: Roll back rule and data instances in a specific application.

For more information, see Using restore points to enable error recovery.

Rules can no longer access Pega internal Java packages

You can no longer create rules that access Java packages that reference internal APIs (syntax com.pega.platform.*.internal*). This change does not affect rules that access Pega public API packages.

If you encounter issues when running existing rules that reference internal Pega APIs, contact Pega Support.

Upgrade impact

After an upgrade to 8.4 and later, clients can no longer save new or modified rules that access Pega internal APIs; existing rules that reference internal APIs can still be run but cannot be modified. 

What steps are required to update the application to be compatible with this change?

Following a software upgrade to 8.4 or later, clients can refactor existing rules into guardrail compliant rules. To find rules to refactor, run the validation tool from designer studio (Application > Tools > Validation) to identify what rules fail validation; failed rules that include the message "Test compilation failed : Illegal internal class reference : com.pega.internal.XYZ" can updated to reference appropriate APIs.

Client-server deployment of Hazelcast

Pega Platform™ now supports the client-server model for the Hazelcast service, which provides cluster communication and distributes Pega Platform features across nodes. This optional deployment model introduces independent scalability for both servers and clients in Pega Platform, improving stability for deployments that use a large number of nodes.

The Hazelcast client-server model is intended for select environments running Pega Platform 8.4 across a large number of nodes and should be deployed only when directed by Global Client Support.

Improved indexing performance by gradual retrieval of data

Search functionality in Pega Platform™ versions from 8.4.5 to 8.6 now includes the option to improve indexing performance when reading query results from a large table in a database. For example, to load the recommended 50 records at a time, in the Pega-SearchEngine ruleset, create the indexing/distributed/fetchsize dynamic system setting, and set the value to 50.

Creating the fetchsize dynamic system setting ensures that your system does not crash while indexing classes with numerous instances.

For more information, see Creating a dynamic system setting.