INC-141579 · Issue 591852
Enhancement added for node heartbeat recovery process
Resolved in Pega Version 8.3.5
Nodes were not showing up in the admin portal even though they were up and running and could be seen in the pr_sys_statusnodes table. The exception "An exception was encountered while invoking the cluster membership listener callback" was seen. All nodes became visible again after multiple restarts. The root cause was traced back to a temporary database connectivity problem. The database itself was fine according to database monitoring reports, but a network problem, a slow database query, or another issue prevented Pega from establishing a connection for more than a minute. An enhancement has been made to resolve this: if a node becomes unhealthy due to the service registry missing due to a failed heartbeat, the heartbeat will try to recover after 60 seconds and keep trying every 30 seconds until it succeeds.
INC-142674 · Issue 594496
Enhancement added for node heartbeat recovery process
Resolved in Pega Version 8.3.5
Nodes were not showing up in the admin portal even though they were up and running and could be seen in the pr_sys_statusnodes table. The exception "An exception was encountered while invoking the cluster membership listener callback" was seen. All nodes became visible again after multiple restarts. The root cause was traced back to a temporary database connectivity problem. The database itself was fine according to database monitoring reports, but a network problem, a slow database query, or another issue prevented Pega from establishing a connection for more than a minute. An enhancement has been made to resolve this: if a node becomes unhealthy due to the service registry missing due to a failed heartbeat, the heartbeat will try to recover after 60 seconds and keep trying every 30 seconds until it succeeds.
INC-137255 · Issue 584910
Resolved Cassandra query read/write latency
Resolved in Pega Version 8.3.5
Read/write latency was seen when certain Cassandra queries were triggered by the system. This was traced to an unusual case where a flush was not done for an exceptionally long time period, and has been resolved.
INC-143121 · Issue 610735
Timeout for loading predictors made configurable
Resolved in Pega Version 8.5.3
When using an extremely large number of predictors, the Report definition pzADMPredictorsFilter was suffering timeouts due to the time for loading predictors from the database exceeding the time threshold allowed. This has been resolved by marking the rule as editable to allow custom setting of the threshold according to need.
INC-150395 · Issue 625070
Tokenizer updated to handle commas
Resolved in Pega Version 8.5.3
The Text Analyzer was not working as expected in cases where the number was combined with a comma (,) with it but was working when a space was used between the number and the comma. This was traced to the tokenizer not correctly processing and splitting the input text when there was a special character before or after the token. This has been resolved by updating the tokenizer logic.
INC-150873 · Issue 612897
Performance improvement for saving ADM model rule
Resolved in Pega Version 8.5.3
Saving an ADM model rule generated a heap dump. The stack trace from the heap dump showed a thread consuming the maximum memory (4.7 GB of memory). Configurations on all factories are updated when a model rule is saved, but at the time of development it was not expected that there would be a lot of factories in a Dev environment so the system was loading all existing factories simultaneously into memory and updating configurations on them. To improve performance, an update has been made which will now sequentially load factories and update the configuration.
INC-151037 · Issue 609626
Enhanced ADM diagnostic logging
Resolved in Pega Version 8.5.3
Diagnostic logging enhancements have been added in order to better identify issues related to ADM models not being created or where learning is impacted.
INC-151421 · Issue 614596
Adaptive Model Save-as works With Predictor Type
Resolved in Pega Version 8.5.3
A predictor change which was accounted for in a previous adaptive model rule version did not work on future save-as. This was traced to a missed use case that when the value 'pyPrevPredTransformation' is empty, validation on that is unnecessary. This has been resolved by adding an 'if' condition in the validation activity.
INC-153223 · Issue 613705
DSS added to set Cassandra query page size limit
Resolved in Pega Version 8.5.3
When a site with a large number of nodes captured responses to the commit log, it was possible for nodes run out of heap space and cause system instability. This was due to the Cassandra query not having a specified page size, so the cluster-wide default setting of 5000 created an issue for large sites which could have over 150 nodes capturing responses. This has been resolved by specifying the page size on the Cassandra query that used to read responses from the commit log. This value will be set dynamically with a DSS to not read more than n responses from all shards combined in a single batch.
INC-154128 · Issue 614297
High thread contention in StrategySSAProgram resolved
Resolved in Pega Version 8.5.3
While Running inbound campaigns during peak times, high contention was observed around "threadSafeWrite" and "threadSafeRead". To resolve this and improve performance, the code around handling write locks has been updated.