Hi everybody ! I must do a test for the 2d level cache.
But I've stange results :

I'm doing an object with read-write strategie cache.

In my code :

Open session 1.
Open session 2.
Session 1 read the object by his id.
Session 2 read the object by his id.
My session1 update the object (and flush it).

My session 2 evict the object, an try to read the object by his id. Normally here the session 2 should found the object in the second level cache, but the traces say that the object is locked and session 2 go directly to database.

Anobody has an idea ?

Thanks in advance, Yann.

PS: excuse my poor enlgish (French).

Sounds like correct behavior to me. Try nonstrict-read-write if you don't care about transaction isolation.

Thanks for your response gavin, but, normally if the session 1 has flushed her object, I think it shouldn't be locked any more, I'm wrong ? And each new session created after this failed to check the object in the 2d level cache until I send a refresh.

If you want to understand how the second-level cache concurrency strategies enforce isolation, it is better to read the code and JavaDoc. I don't have time to explain it here.

People don't seem to appreciate just how sophisticated Hibernate's cache algorithms are, nor how difficult a problem caching is, in general.