Support Article

Getting lot of alerts from FTSIncrementalIndexer agent

SA-26561

Summary



Getting lot of alerts from FTSIncrementalIndexer agent causing heap size issue and node crash.

Error Messages



2016-06-20 10:19:07,245 [fault (self-tuning)'] [ STANDARD] [GSSC_Agents:01.01.01] ( internal.async.AgentQueue) ERROR - Problem queue Pega-SearchEngine #0: System-Queue-FTSIncrementalIndexer.pzFTSIncrementalIndexer will restart in 240000 ms
2016-06-20 10:38:26,410 [b7e5.cached.thread-6] [ STANDARD] [ ] ( cluster.impl.TcpIpJoiner) FATAL - [--]:6031 [f124e363b417e752c711b0953b55b7e5] [3.4.1] Java heap space
java.lang.OutOfMemoryError: Java heap space
2016-06-20 10:44:38,315 [7e5.cached.thread-13] [ STANDARD] [ ] (zelcast.cluster.ClusterService) FATAL - [--]:6031 [f124e363b417e752c711b0953b55b7e5] [3.4.1] Java heap space
java.lang.OutOfMemoryError: Java heap space
2016-06-20 10:44:52,202 [c-operation.thread-3] [ STANDARD] [ ] (ns.MasterConfirmationOperation) FATAL - [--]:6031 [f124e363b417e752c711b0953b55b7e5] [3.4.1] Java heap space
java.lang.OutOfMemoryError: Java heap space


Steps to Reproduce



Not Applicable 

Root Cause



A defect in Pegasystems’ code or rules. The agent name is being sent as null to the stored procedure "sppr_sys_reservequeueitem_b" while dequeuing during incremental indexing.
In DistributedIndexerAgent and SystemWorkIndexerAgent, where we are getting the queueIterator object we are only setting the className and not sending the appropriate agent name and hence a null value is being sent from within the engine to this stored procedure.


Resolution



Sending the agentName as well with the class name when retrieving an instance of queueIterator .

HFix-28521 addresses the issue.
 

Published August 8, 2016 - Updated August 12, 2016

Have a question? Get answers now.

Visit the Collaboration Center to ask questions, engage in discussions, share ideas, and help others.