Hi All,

I'm using search 3.1.0.

I have a object called Feed which represents an RSS feed and an object called News which represents a RSS feed item.

Data is indexed in lucene, in a way where the keywords of the RSS feed are also indexed as part of the News, so that when the user enters some keywords that appear in the RSS feed description, they would get all data possible including the RSS feed and the news related to it.

So basically I use @ContainedIn and @IndexedEmbeded on the references to make that happen.

Sometimes I update a field on the Feed object called "state". This field is not indexed. Note to generalize the matter lets say the problem refers to all non indexed fields being modified.

My problem is whenever it is changed, Hibernate search removes the feed and all news items from the index and then reindexes them.

The real cost of the operation was to change a byte in the DBMS now we:
1. Fetch the news collection every time the state is modified.
2. Reindexing every news item, and the feed itself including analisys, term creation, IO etc.
3. Perform needless IO operations.
4. Potentially de-optimize the index, so now search is slower.
5. static filters and caches have to be re-generated due to the indexreader being out of date.

I have read some posts about this on the forum, but none of the answers seem to deal with the issue.

Does anyone have an idea on how to solve this problem?

Not really possible today but I think solving
http://opensource.atlassian.com/project ... SEARCH-361
would do what you want. that's an easy patch btw if someone is interested to contribute.

sure post the patch to the bug please