Passivation should definitely not kill a stateful session bean. (Does JBoss
do this? So far, I haven't deployed anything in JBoss that is idle for that
long a period.) However, I wouldn't be surprised if JBoss periodically
aged-out unused stateful session beans. I think that most good app servers
will do this to clean up after clients which have died, or have not politely
disconnected.
If that's what's happening, the ideal option would be to increase the max
bean age to accomodate your longest expected login period, but leave
passivation as is. If that's not what's happening, then maybe it's a bug?
Mike
----- Original Message -----
From: "Mike Finn" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Sunday, July 15, 2001 7:33 PM
Subject: Re: [JBoss-user] SB timeout -> passivation -> activation failure
> Thanks Michael & Burkhard
>
> > I haven't done much experimenting with performance settings, but you
might
> > be able fix this with some adjustments the container cache policy. See
> > http://www.jboss.org/documentation/HTML/ch06s08.html, the section on
> > "Advanced cache configuration". You might try adjusting the
<max-bean-age>
> > setting. (I assume this can apply to stateful session bean timeouts.)...
> > Michael
>
> This might solve an immediate problem but won't do in the long run see
> below...
>
> > if this id SLSB you have to tweak your container settings, as Mike said,
> > but with SFSB, you should be able to reactive your session?!?
> > SL=stateless, SF=statefull, BUT I'm not 100% sure, since I would call
this
> > a design issue having sessions of several hours, there SHOULD be other
> > ways... Burkhard
>
> Maybe a bit more info will help. I am building the server side with EJB's
> because I very much like the promise of putting business login in a middle
> tier (in this case an enterprise java app server ie JBoss).
>
> Phase one of my project is to build a somewhat traditional client app
(Swing
> in this case) that uses the services of the app server instead of the
2-tier
> model which invariably would result in some business logic in the client
app
> (yuck - been doing that for far too long). So an office worker using the
app
> needs to authenticate once in the morning and have access all day long...
> ie. a long stateful session. Passivation is okay (in fact fantastic) but
> only if (re)activation works.
>
> Phase two of the project is to open up access up to business partners and
> customers via web or tiny-apps. These will also use session beans
(hopefully
> the same one as the former) but will be much shorter lived.
>
> So, I am happy to change the bean aging, and even clone the bean so that I
> can have one that is short-lived and one that is long-lived but the long
> lived one must be able to be able to survive multiple
passivation-activation
> cyctes until it is cleanly terminated by the client or a suitably long (in
> the order of hours) timeout occurs.
>
> Right now it seems once passivation occurs on the stateful bean the
session
> dissapears into a black hole, and when the server tries to activate it it
> blows. Surely this is not correct behaviour on part of the server? And
this
> is what I am trying to fix.
>
> Sorry for the long winded answer but figured more was better than not
enough.
>
> Thanks
> Mike.
>
> ===================
> Mike Finn
> Tactical Executive Systems
> [EMAIL PROTECTED]
>
> _______________________________________________
> JBoss-user mailing list
> [EMAIL PROTECTED]
> http://lists.sourceforge.net/lists/listinfo/jboss-user
>
_______________________________________________
JBoss-user mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/jboss-user
