Entity Beans and row-locking

[morenito9000]

Hi all,
I have 2 clients that call a stateful session bean's method.

Inside this method the session bean starts a transaction and
then retrieves/update an entity bean.

The behaviour of the method differs depending
the client that called it.

In particular, for Client 1 the method performs this steps:

1) retrieves the entity bean (with findByPrimaryKey())
2) sleeps for 45 seconds
3) updates the entity bean

For Client 2 the steps are:

1) retrieves the entity bean (with findByPrimaryKey())
2) updates IMMEDIATELY the entity bean

I'm doing some tests about competion among these 2 clients.
I run Client 1 and after few seconds I run Client 2.

I expect that Client 2 ends before Client 1 but the strange
thing is that Client 2 waits till the end of the sleep(45) of
Client 1 !!

I'm non able to understand this behaviour :-(

I'd have understood it if there had been a
<row-locking> tag with TRUE value in the
defaults section of standardjbosscmp-jdbc.xml file,
but this tag is set to FALSE.

<defaults>
 <datasource>java:/PostgresDS</datasource>
 <datasource-mapping>PostgreSQL</datasource-mapping>

 <create-table>false</create-table>
 <remove-table>false</remove-table>
 <read-only>false</read-only>
 <read-time-out>300000</read-time-out>
 <row-locking>false</row-locking>

Can anyone give me an explanation ?

Many thanks
Moreno

[aloubyansky]

This is because of pessimistic locking. Check the JBossCMP wiki Locking page.

[morenito9000]

Thank you Loubyansky !

I read the JBoss book's pages concerning Optimistic-Locking.

If I wanna Optimistic-Locking I must use, for my entity bean,
a Container-Configuration like 'Instance per Transaction...."
and define the fields in my bean that I'll use for optimistic-locking policy
(I hope I understood well :-))

Just only two questions:
1) if I declare to use Optmistic-Locking
it has no sense to set the row-locking tag to TRUE ?

2) And if I don't wanna Optimistic-Locking I must do
nothing (that is the default behaviour)
like I experimented with the "two beans" test ?

Thanks again
Moreno