AnsweredAssumed Answered

Alfresco2.1 index purge problem

Question asked by ngchiwing on Aug 1, 2012
Hi Folks,

I’m new here in the forum, and tackling a problem in indexes purging.

We are using Alfresco 2.1 on Fedora14 with about 3 million transactions and 2 million nodes. Recently, due to performance and storage space issue, we tried to purge certain amount of contents and they decently go into the “alf_data/contentstore.deleted” directory. However, we found that the size of the “alf_data/lucene-indexes” directory still huge and didn’t get any smaller.

So, we replicated the system into a testing machine and have tried some methods to rebuild the indexes (hope that the rebuilt indexes will release some space), but failed :-

Method 1 : set “index.recovery.mode” to “FULL” in file “”
-   set “index.recovery.mode” to “FULL” in file “”
-   remove the “lucene-indexes” directory
-   We waited the index recovery during system restart for about 3hrs, and not even 10% completed. Also, the size of “lucene-indexes” directory didn’t grow. So, we stopped it.

Method 2 : add “BAHFullIndexRecoveryComponent.class” into “alfresco-repository.jar”
-   according to we have added the “BAHFullIndexRecoveryComponent.class” into the “alfresco-repository.jar”
-   add related setting into the “index-recovery-context.xml” file
-   set “index.recovery.mode” to “FULL” in file “”
-   remove the “lucene-indexes” directory
-   this time we have more patient, we waited for about 24hrs or more, and we saw the size growth of the “lucene-indexes” directory. However, it finally came to error as :
ERROR [org.hibernate.transaction.JDBCTransaction] JDBC rollback failed
com.mysql.jdbc.exceptions.MySQLNonTransientConnectionException: Connection.close() has already been called.
Invalid operation in this state.

At the end according to , we listed out the 10 indexes directories from the file “IndexInfo” inside the “lucene-indexes/workspace/SpacesStore” directory. We deleted all other directories inside the “lucene-indexes/workspace/SpacesStore” except the 10 directories described by the “IndexInfo” file and the “IndexInfo” file itself. Then we restarted the system, it worked like normal and seems that the originally contents inside the system still there. We have tried a lot of file add / delete / search operations, and it still worked like normal.

So, the question is: can we use it as a solution for index purge in the system? Any advice would be appreciated. Thanks in advance!

Best Regards,