Batch Data Flow causes CPU starvation when errors
User is having Pega 7.1.8 running with Apache Tomcat v6.0.28 on a local Windows OS and connecting to Oracle 11g. And, user is having Sun JDK 1.6.0_21 installed on this system.
The problem occurs each time an error is encountered and this is the observed behavior:
(1) The CPU utilization increases dramatically.
(2) The time to process is extended by a considerable margin for each error that is encountered.
(3) On occasions even at modest volume the environment (node) has been compromised. This has happened in both a local instance and a cloud instance in use for development.
None. Specifically the error is reported via the Data Flow run control panel and following exception was reported:
Caused by: com.pega.pegarules.pub.PRRuntimeException: Strategy component Strategy___Result___Final___Output failed to execute: Strategy component WEBDataCollection failed to execute: PropertyValueInvalid null .pegaToDouble() Invalid number format (was 1.0)
... 138 more
Caused by: com.pega.pegarules.pub.PRRuntimeException: Strategy component WEBDataCollection failed to execute: PropertyValueInvalid null .pegaToDouble() Invalid number format (was 1.0)
... 140 more
Steps to Reproduce
Execute any Data Flow to process a batch of input data calling the decision logic (Strategy) that results in an error being detected that is a PRError being reported in the PRPC log file.
In Pega 7.1.8, following were the two issues:
1: In case decision Strategy contains lots of components, the java call stack is too deep and it takes a lot of time to generate and save it.
2: The "1.0" NumberFormatException.
0% found this useful