From reading the documentation and the hibernate in action book (page 165 - setting an isolation level) my understanding is that the isolation level cannot be changed for conenctions obtained from a datasource in a managed environment.

We are using Websphere 5.1, Session Beans calling hibernate code via Spring, conenction to DB2 UDB for iSeries

What we are finding is that in a multi-user test, the whole files are getting locked - this to me is consistent with an isolation level of read uncommited.

When setting up an iSeries datasource I can't see an option to set a default transaction isolation level - when writing manual SQL stuff we have always set the isolation level on the session bean.

Can someone clarify where I should be setting the isolation level so that the default is repeatable read ? Is it in the hibernate properties file, on the session bean, or on the datasource ?

Sorry, I answered my own question - it was a Monday morning.....

You set the "transaction isolation" property on the datasource to be "repeatable read".

Hi Paul, I am having was I believe is a concurrency problem with my Java app running on an iSeries. I am seeing SQL errros (SQL99999) and when I do my research, this error can happen when my isolation level is not set correctly and/or my autocommit is set on. My issue is that our physical files on the iSeries are NOT journaled. That means I can not have transactions. This limits me in what I can do. Would you have any suggestions how to get around this type of issue?

Thanks

There is no way you can have transactions (well on DB2 for iSeries at least, who knows for other things), without having your physical files journalled. This is because the transaction has to do eitehr a commit or a rollback depending on whether it worked or not, the job has to be running under commitment control (you don't have to worry about that, the JT400 JDBC driver handles all that for you).

Setting up journalling on tables is a very trivial task - what's the issue with them not being journalled ?

This may be teaching you to suck eggs, but if you want to journal them, creat a journal receiver using CRTJRNRCV, create a journal using CRTJRN, specifying the journal receiver you just created, then start journalling on each table via the STRJRNPF command. If you set up a PDM option to start journalling that will make things easier.