SR-D20439 · Issue 496405
Hotfix hashmap cleanup improved
Resolved in Pega Version 8.1.7
A DL installation failed with an out-of-memory error. It was observed that many CacheEntry(VersionedJdbcJarRwader.java) hashmaps were being created to store the details for each hotfix but that were then never cleared. This has been resolved by adding the finally block in CodeImportProcessImpl.java class to call the shutdown method for proper cleanup. In addition, the default command line utility (prpcUtils) has been updated to use JVM settings of Xmx 4GB and capture a heap dump if an out-of-memory error is encountered. The auto-generated prconfig.xml has also been updated to leverage its minimal startup setting which will avoid loading the conclusion cache into memory.
SR-D28538 · Issue 502059
Corrected requestor status flag for direct map
Resolved in Pega Version 8.1.7
Numerous "Unable to create requestor" alerts were logged. This was traced to an error in HttpAPI where after retrieving the requestor from the internal requestor map directly, the requestor creation status flag was not set properly. This caused last action to post that alert instead of the correct notice of "existing requestor retrieved". This has been resolved so the flag reflects the correct status.
SR-D51554 · Issue 514064
Local UUID cache will be updated when merge event is detected
Resolved in Pega Version 8.1.7
Cluster-related issues were seen in multiple production clusters. For some nodes in the cluster the Cluster Management screen showed all expected nodes with valid Node IDs displayed, and on other nodes the Cluster Management screen showed the node ID of itself, SERVER@localhost:5701. On an impacted node displaying the wrong ID, the Node Information landing page did not work and displayed the error "Unable to execute job on ." Multiple advanced agents running on nodes in the affected clusters, both with correct and incorrect IDs, also failed with a similar error "Unable to execute job on <node's job id>". This was traced to a merge performed after a split brain. To resolve this, the code has been updated to handle merge events: when the node UUID is changed as part of a split brain recovery, the local UUID cache will be updated when the merge event is detected.
SR-D20423 · Issue 503447
Improved upgrade handling for tables using classes with property references
Resolved in Pega Version 8.1.7
After upgrade, some page property values were blank and exposed database columns did not contain the values. Investigation showed that the reference properties did not have context and hence column population was not able to determine their value. To resolve this, instead of doing column population for all the columns, the system will identify classes that have property references and that are being optimized as part of upgrade process and maintain a list of specific columns to update.
SR-D29485 · Issue 503514
Enhancement added to modify URL encryption for load testing
Resolved in Pega Version 8.1.7
An enhancement has been added which allows conditionally modifying URL encryption for load testing. This uses the flag crypto/useportablecipherforurlencryption: if true, a portable hardcoded key is used to encrypt the URLs and if false, a dynamically generated key per thread/requestor is used to encrypt the URL.
SR-D52969 · Issue 514705
Column population honors thread count of 1
Resolved in Pega Version 8.1.7
The thread count parameter in the column population activity was not being honored, causing repeated deadlocks when trying to populate columns. Investigation showed that the ExposeCols process did not honor the thread count when it was 1 (the default is 4), and this has been fixed by adding the necessary code so that if the thread count is 1, it will not run in multhreaded mode.
INC-145293 · Issue 610935
Additional diagnostic logging added for ElasticSearch startup issues
Resolved in Pega Version 8.5.6
The PyIndexerState was stuck in Starting status during node initialization. This issue could occur if the filesystem became hung due to network level issue while scanning entries from /etc/mtab, resulting in a lock which was not released correctly. In order to better determine which node entry in a cluster may be responsible for the hang, an update has been made which will use a temporary virtual environment to repeat the part of the initialization phase responsible and generate additional logs for debugging. To activate this, the PegaSearch.Diagnostics logger must be set in DEBUG mode. This duplicated virtual initialization will not interrupt the normal initialization.
INC-163791 · Issue 666195
Simplified default reference time calculations
Resolved in Pega Version 8.5.6
After a job scheduler was configured to run at Start time = 21:00:00 for Time zone = Europe/London, the scheduler determined 20:00:00 as the next start time. This was due to the calculation for the next start time using the time zone offset calculation pattern for the date and time stored in System-Runtime-Context.pxCreateDateTime, which had difficulty with changes to the time zone definition implemented in the time between the given date and today (meaning the current time) such as daylight savings time. To resolve this, the default reference time from System Runtime Context will be 'now' instead of Date(0).
INC-168253 · Issue 660882
Documentation updated for using Connect MQ with clustered queues
Resolved in Pega Version 8.5.6
After update to Pega Platform v8.6, an issue was seen with Connect MQ rules that used an alias queue. The support documentation has been revised to outline the necessary steps for updating an application to perform as expected. In the Pega-IntegrationEngine ruleset, create the following dynamic system settings and their default values: "mq/pmo" = 0 "mq/gmo" = 0 "mq/oo_request" = 1040 "mq/oo_response" = 16 For more information, dynamic system setting support articles are available.
INC-174625 · Issue 655241
Admin Studio will consider cluster protocol when returning listener status
Resolved in Pega Version 8.5.6
When using a few nodes in standalone mode for BIX extract combined with server nodes using Hazelcast, opening the admin studio pages with service discovery caused an error to be thrown. This was traced to the system writing an entry to pr_sys_statusnodes table as an embedded node whenever a BIX extract was triggered, causing those standalone nodes to be incorrectly considered by the listener landing page. This has been resolved by configuring the system to either return the local member when the cluster protocol is standalone or to return all Hazelcast members if the cluster protocol is Hazelcast.