Alfresco 2 - Reindex Full vey slow

[stylus753]

Configuration
- SunFire V240 with 4GB RAM and SunOS 5.10
- N° Documents 167.602
- dir.root : 18 GB occuped , disk free 60%
- dir.contentstore : 35 GB occuped,  disk free 80%
- Alfresco Enterprise ver.2.2.0
- Jboss Portal ver.2.6.5
- Oracle 10g rel.2

Hi,
after the problem with search fails i set the index recovery mode FULL then restart but this process it's very very slow and terminated with timeout Jdbc after 14Hours
each step of process rebuild is bigger then 2 Hour

18:53:27,313 INFO  [org.alfresco.repo.node.index.FullIndexRecoveryComponent] Index recovery started: 167,880 transactions.
20:00:25,858 INFO  [org.alfresco.repo.node.index.FullIndexRecoveryComponent]    10 % complete.
21:10:11,334 INFO  [org.alfresco.repo.node.index.FullIndexRecoveryComponent]    20 % complete.
23:05:49,801 INFO  [org.alfresco.repo.node.index.FullIndexRecoveryComponent]    30 % complete.
01:36:01,231 INFO  [org.alfresco.repo.node.index.FullIndexRecoveryComponent]    40 % complete.
03:05:07,831 INFO  [org.alfresco.repo.node.index.FullIndexRecoveryComponent]    50 % complete.
05:00:54,973 INFO  [org.alfresco.repo.node.index.FullIndexRecoveryComponent]    60 % complete.
07:46:19,514 INFO  [org.alfresco.repo.node.index.FullIndexRecoveryComponent]    70 % complete.
09:57:22,152 INFO  [org.alfresco.repo.node.index.FullIndexRecoveryComponent]    80 % complete.
11:02:02,044 ERROR [org.hibernate.transaction.JDBCTransaction] JDBC rollback failed
java.sql.SQLException: Closed Connection
        at oracle.jdbc.driver.DatabaseError.throwSqlException(DatabaseError.java:112)
        at oracle.jdbc.driver.DatabaseError.throwSqlException(DatabaseError.java:146)
        at oracle.jdbc.driver.DatabaseError.throwSqlException(DatabaseError.java:208)
        at oracle.jdbc.driver.PhysicalConnection.rollback(PhysicalConnection.java:1142)
        at org.apache.commons.dbcp.DelegatingConnection.rollback(DelegatingConnection.java:265)
        at org.apache.commons.dbcp.PoolingDataSource$PoolGuardConnectionWrapper.rollback(PoolingDataSource.java:288)
        at org.hibernate.transaction.JDBCTransaction.rollbackAndResetAutoCommit(JDBCTransaction.java:183)
        at org.hibernate.transaction.JDBCTransaction.rollback(JDBCTransaction.java:162)
        at org.springframework.orm.hibernate3.HibernateTransactionManager.doRollback(HibernateTransactionManager.java:581)
        at org.springframework.transaction.support.AbstractPlatformTransactionManager.processRollback(AbstractPlatformTransactionManager.java:745)
        at org.springframework.transaction.support.AbstractPlatformTransactionManager.rollback(AbstractPlatformTransactionManager.java:722)
        at org.springframework.transaction.interceptor.TransactionAspectSupport.completeTransactionAfterThrowing(TransactionAspectSupport.java:325)
        at org.alfresco.util.transaction.SpringAwareUserTransaction.rollback(SpringAwareUserTransaction.java:498)
        at org.alfresco.repo.transaction.RetryingTransactionHelper.doInTransaction(RetryingTransactionHelper.java:309)
        at org.alfresco.repo.transaction.RetryingTransactionHelper.doInTransaction(RetryingTransactionHelper.java:190)
        at org.alfresco.repo.node.index.AbstractReindexComponent.reindex(AbstractReindexComponent.java:220)
        at org.alfresco.repo.node.index.IndexRecoveryBootstrapBean.onBootstrap(IndexRecoveryBootstrapBean.java:35)
        at org.alfresco.util.AbstractLifecycleBean.onApplicationEvent(AbstractLifecycleBean.java:62)‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍

a) How set the alfresco/lucene parameter, and alfresco/hibernate transaction,  for optimize the reindex process?
b) It's possibile to start alfresco api for only reindex document with no start alfresco.war include in the container JEE (in this case Jboss)?

thanks in advanced

[andy]

Hi

Unless this is all AVM data this should not take that long - even with your setup.
I do not think 2.2 had multi-threaded index rebuild.

The V240 was EOL in Sept 2006 🙂

Did you delete the index first?

I would suggest you upgrade or change your DB timeout, ….

Andy

[stylus753]

in dir.root the directory is
audit.contentstore, backup-lucene-indexes, contentstore.deleted and lucene-indexes
therefore i delete the directory "lucene-indexes" and restart the index Full?  :roll:

[andy]

Hi

yes. Do not delete the content stores.
The rebuiild will be faster - it may still time out - so up the DB timeout.

Andy