Hi,
After fixing a bug in my test case I can see that this is now indeed
fixed.
However, I'm going to be a pain here. I don't think that your test case
code in
org.jboss.test.JBossTestCase.testLocalInterfacePassivation
and
org.jboss.test.cts.ejb.StatefulSessionBean
proves anything. It only proves that at some point one or both of
ejbPassivate/ejbActivate have been called. Before the bug was fixed, the
failure was occurring after ejbPassivate was called.
In order to properly test that the StatefulSessionBean was passivated
and then recreated you need a transient flag that gets set in
ejbPassivate. Your non-transient 'wasPassivated' flag will itself be
serialised and then reinstated at activation time, but for good measure,
you set it again in ejbActivate anyway. Consequently, you can't tell if
the bean was ever successfully passivated and then reactivated. The only
way to tell is by exploiting the fact that transient members aren't
reinstated when the bean is reactivated.
Therefore, I think that:
1. org.jboss.test.cts.ejb.StatefulSessionBean.wasPassivated should be
transient (and probably renamed - that's why it's just called
'passivationFlag' in my test case).
2. the line containing 'wasPassivated = true;' should be removed from
ejbActivate().
For a slightly stronger test, leave the 'wasPassivated' flag as is, and
add a second transient field so that you can check that passivation was
actually attempted and it succeeded (because the transient field gets
reset).
Steve Coy
On Friday, April 19, 2002, at 02:58 AM, [EMAIL PROTECTED] wrote:
> Bugs item #541855, was opened at 2002-04-09 22:21
> You can respond by visiting:
> http://sourceforge.net/tracker/?func=detail&atid=376685&aid=541855&group_id=
> 22866
>
> Category: JBossServer
> Group: v3.0 Rabbit Hole
>> Status: Closed
>> Resolution: Fixed
> Priority: 6
> Submitted By: Stephen Coy (scoy)
> Assigned to: Scott M Stark (starksm)
> Summary: Passivation of stateful sess beans fails
>
> Initial Comment:
> Section 7.4.1 of the EJB2.0 spec says that
> EJBLocalHome and EJBLocalObject (amongst
> others) should be persisted during passivation.
> The code in
> org.jboss.ejb.plugins.SessionObjectOutputStream
> clearly doesn't do this yet.
>
> Someone needs to check and ifx this code for
> EJB2.0 conformance.
>
>
> ----------------------------------------------------------------------
>
>> Comment By: Scott M Stark (starksm)
> Date: 2002-04-18 09:58
>
> Message:
> Logged In: YES
> user_id=175228
>
> Fixed on head and the 3.0 branch
>
> ----------------------------------------------------------------------
>
> Comment By: Stephen Coy (scoy)
> Date: 2002-04-17 01:10
>
> Message:
> Logged In: YES
> user_id=463096
>
> Doh! I figured out that I can set the cache size to 1.
> Creating a second session bean causes the first to be
> passivated. We now have a test that successfully fails.
>
> ----------------------------------------------------------------------
>
> Comment By: Stephen Coy (scoy)
> Date: 2002-04-17 00:17
>
> Message:
> Logged In: YES
> user_id=463096
>
> The attached files incorporate test case "stateful2" into
> the JBoss test suite. Please remove any .DS_Store
> files you find - they're an unfortunate side effect of using
> MacOS at present.
>
> I spent quite some time trrying to configure the
> container so that the session bean would passivate
> before the test completed, but failed miserably.
> However, you can still see the passivation failing when
> the EJB is undeployed.
>
>
> ----------------------------------------------------------------------
>
> You can respond by visiting:
> http://sourceforge.net/tracker/?func=detail&atid=376685&aid=541855&group_id=
> 22866
>
> _______________________________________________
> Jboss-development mailing list
> [EMAIL PROTECTED]
> https://lists.sourceforge.net/lists/listinfo/jboss-development
_______________________________________________
Jboss-development mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/jboss-development
