Process Portal: Troubleshooting problems with search results

After users run a search in Process Portal, they receive error messages that indicate too many results exist for the search criteria, and that the search must be refined.

Symptom and cause

Process Portal users receive the following error messages after they run searches:
  • In the Process Performance dashboard: The search result is too large. Refine your search and try again.
  • In the Work or saved search dashboards: CWTBG0567E:Too many potential search results. To limit the number of search results, use more specific search terms or AND operators with the search terms.
In the trace file, you might also see the following stack information:
[99-9-99 13:03:04:354 CST] 00000064 TaskSearchRet 1 com.ibm.bpm.search.task.search.TaskSearchRetrievalImpl searchTaskIndex
Task Search Found 108601                                                
[99-9-99 13:03:04:354 CST] 00000064 TasksResource 2                    
com.ibm.bpm.rest.impl.task.TasksResource getWleQueryResult Found too    
many hits, exceeding the allowed maximum of 100000

In both error situations, the search results are limited by the value of the <index-search-max-hits> property in the 99Local.xml server configuration file. The default value of the property is 100000. The limit exists to avoid server-side performance issues that can happen with runaway queries. Runaway queries can happen, for example, if both the full-text search and saved search result sets are large. In the trace file example, the full-text search returns 108601 results, which is more than the configured default limit of 100000.

Resolving the problem

  • Process Portal users can refine their searches by including filters in the search criteria.
  • Administrators can override the default value by increasing the value of the <index-search-max-hits> property in the 100Custom.xml file. Be aware that higher property values impact response time and memory usage. Therefore, before you increase the value of the property in production environments, test the effect of the higher search limit on performance.