This happens on hibernate 3.2.6, and older versions as well

When running on Oracle9Dialect, with an Oracle10G DB, Hibernate generates UPDATE statements in cases where we simply send SELECT and SELECT count() statements. This only happens on Oracle. MySQL does not do that. I can provide the generated SQL statements, but before I do that (they are huge), is there an obvious answer to that?

This is causing us lock problems betwen two transacted services, where I feel it should not. One of the transacted call is a read only call that simply does two selects: One select to return paginated rows, and another select that returns the count of all rows.

--
Mel T

Hi. I found the same problem using Castle Project RC3 (nhibernate build 1.2.0.4000?) and IBM DB2 UDB running on iSeries/AS400.

In fact this update sentences are causing locking problems when I'm just trying to read data, no need to update at all.

I tried to change isolation levels but the problem remains the same.

Just ask if you need more info (generated sql sentences, logs, whatever).

This is a big issue for my application. Hope we get an answer. Thanks!

We eventually fixed the problem. There were calls on our service layer that were transacted (default settings of the transaction annotation) when they should not have been. And because the entities in these calls were eventually updated inside the service, even though we did not call explicitly "update" on the entity manager, hibernate noticed the change and updated the DB. That had gone unnoticed to me.

A cleanup of the transaction declarations fixed the problem.

deleted duplicate reply