Need help with Hibernate? Read this first:
http://www.hibernate.org/ForumMailingli ... AskForHelp

Hibernate version:
3.3.0 GA
hibernate search 3.1.0 beta

Hi i was wondering whats the fastest way to perform weekly batch process where you do manual batch processing (i read the guide). It seems it would delete all old index and reindex again. I want to be able to index the same thing again through a periodical update (like once every three to five days, or week, etc).

Hi,

you could implement a TimerTask or your a scheduling library like quartz to configure your batch indexing.

Why do you want to run this batch indexing in a scheduled way. Once you have an initial index it should stay in sync due to Hibernate Search transparent index synchronization.

--Hardy

in order to keep the domain object clean to be exported as java rmi services.

Hi,

I am not quite sure what this has to do with RMI. Can you maybe elaborate on your whole use case?

--Hardy

1.) read from a third party database.

2.) index them.

3.) repeat every two days.

Ok now I understand, at least I think so.
You are indexing and searching against a third part database. You are not modifying this database, but the third party might. Well, in this case using some sort of timer to indx seems to be the best approach. Whether you have to reindex everything in this case depends on whether you can detect via a SQL/Hibernate query which entities have changed (eg last update time stamp). If you cannot determine the changes via a query you have to reindex from scratch.

--Hardy

Ok now I understand, at least I think so.
You are indexing and searching against a third part database. You are not modifying this database, but the third party might. Well, in this case using some sort of timer to indx seems to be the best approach. Whether you have to reindex everything in this case depends on whether you can detect via a SQL/Hibernate query which entities have changed (eg last update time stamp). If you cannot determine the changes via a query you have to reindex from scratch.

--Hardy

Hello,

I'm having a similar problem. I also want to update the index via a TimerTask because I get updates of my database from a third party. So I can not benefit of the Hibernate Search transparent index synchronization.
And I can not determine which entities have changed.

So I tried the folowing:



This method is called e.g. every week.
But this method increases the file size of my index even if there are no changes in the database.
Do I have to remove the old Entities from the index before I force the reindex process with the index(Object o) method? Or is there a different approach to do it?

PS: I'm using Hibernate Search 3.0.1 GA

Thanks for your help!

Hi,

Have you tried optimizing the index after re-indexing? In Lucene documents cannot be just changed. Effectively documents gets deleted and readded with the new values. This fragments the index and hence you should periodically optimize the index. If you rely on the automatic index synchronisation, optimisation is configurable and taken care of. If you index maually you should also optimize the index afterwards.

--Hardy

Thanks a lot Hardy,

after optimizing the index everything is fine. And the file size will not increase anymore if there are no changes in the database.

I thought that the automatic optimization of Hibernate Search has some default value and I whould not have to configure it explicitly.

Thanks a lot for your help!