Question
GE
IN
Last activity: 30 Aug 2017 0:48 EDT
Search Fails
Search in PEGA 7.1.9 is failing. That does not mean search is not working. The search is working but few cases miss out from search. Logs do not have any entries. How do we monitor what has gone wrong with search and remediate it ?
***Edited by Moderator Marissa to update categories***
GE
IN
YEs the location looks fine. The search is not failing completely but few records dont come up randomly. That number of complaints that fail in such a way sometimes is high. How to find where the issue is ?
Pegasystems Inc.
IN
You can try truncating pr_sys_statusnodes and pr_sys_queue_ftsindexer table and then perform reindexing as there are scenarios when some stale data will be present in these tables which might be causing the issue with the searching.
NOTE: But before truncating these tables take the backup of these tables.
J.P. Morgan Services India PVT Ltd
IN
Hi Bargavi,
Are these work objects which are getting failed?
Are the records which are getting failed or you are not able to search are having some similarity in them? Are the records(work object) having some attachments?
Or may be the work object are corrupted some how.?
Pegasystems Inc.
IN
Hi Bargavi
Please go through the below link. Which may helps for further debugging.
https://collaborate.pega.com/question/search-indexing-agent-not-index-few-work-items
Regards
Saran
J.P. Morgan Services India PVT Ltd
IN
I will also suggest you to go through the below video for reference:
https://collaborate.pega.com/discussion/video-support-guide-how-troubleshoot-re-indexing-issues
Thanks
Pegasystems Inc.
GB
If after going through the suggested PDN links etc and depending on how deep you need to go into this: you can directly interrogate the Search Indexes (which are Lucene-based) using various tools.
For instance: this 'Luke' Tool can be used to look directly at the underlying Indexes (you can just copy the underlying Lucene directory to a separate directory - or indeed access them directly - but be careful about modifying them by mistake).
https://github.com/tarzanek/luke
You'll need to build the tool using a combination of 'git' (to fetch the source-code) and 'maven' - to build the JAR file; some of the older binary version of 'Luke' will NOT work with the Elastic Search built Lucene Indexes - due to version differences. (You need Lucene V5 [I think, it might be V6 even]).