IBM Connections HOMEPAGE.SR INDEX_DOCS table size
The HOMEPAGE.SR_INDEX_DOCS table started growing by one million kb a day. How can this growth be stopped?
Index table grows exponentially daily
Indexing tasks are still running on nodes that are no longer in use.
IBM Connections 5.5CR1, Windows
Diagnosing the problem
The steps from "SR_INDEX_DOCS table in Homepage database grows unexpectedly"
http://www-01.ibm.com/support/docview.wss?uid=swg21700854 were implemented but the issue persists.
A review of the SystemOut.log showed the following errors:
[7/20/16 8:09:09:838 CDT] 00000123 IndexCacheMan E com.ibm.connections.search.service.index.cache.impl.IndexCacheManager updateResumePoint CLFRW0261E: An error has occurred in the index manager.
org.springframework.jdbc.UncategorizedSQLException: SqlMapClient operation; uncategorized SQLException for SQL ; SQL state [null]; error code ; Error: executeQueryForObject returned too many results.; nested exception is java.sql.SQLException: Error: executeQueryForObject returned too many results.
[7/20/16 8:09:09:841 CDT] 00000123 DatabaseResum E com.ibm.connections.search.index.process.incremental.DatabaseResumePoint Manager updateResumePoints CLFRW0256E: An error occurred in the Index cache manager.
Confirm all nodes being indexed are actually still in use. This can be done by running the following "SELECT * FROM HOMEPAGE.SR_INDEX_MANAGEMENT" statements then review the output:
SELECT COUNT(*) FROM HOMEPAGE.SR_INDEX_DOCS
SELECT COUNT(*) FROM HOMEPAGE.SR_INDEX_DOCS WHERE FILESCONTENT_ID ISNOT NULL
SELECT COUNT(*) FROM HOMEPAGE.SR_INDEX_DOCS WHERE RESUME_POINT IS NULL
SELECT DISTINCT(CRAWLING_VERSION) FROM HOMEPAGE.SR_INDEX_DOCS
SELECT crawling_version,service,count(*) AS NUM_UPDATE_ITEMS FROM
HOMEPAGE.SR_INDEX_DOCS GROUP BY CRAWLING_VERSION,SERVICE ORDER BY
SELECT * FROM HOMEPAGE.SR_INDEX_MANAGEMENT
SELECT * FROM HOMEPAGE.SR_TASKSDEF
SELECT * FROM HOMEPAGE.LOTUSCONNECTIONSTASK
Resolving the problem
The select statement results confirmed a node no longer in service, was still registered to the indexing task. This caused the index table size to grow quickly.
Once the unused node was identified, the steps from IBM Knowledge Center article "Removing a node from the index management table" were implemented. These steps stopped the table growth and allowed the size to decrease.
Translate this page: