Hey,
Haven't really been reading up on *EVERY* jboss-user email, so don't know if
this has been reported and or fixed in the current repository:
Was using jBoss 2.0 BETA PROD 01 with little problems, and switched
to PROD 02 on Friday. Here is my problem:
I have a simple TimingTest Entity Bean which I have created to record how
fast creating / finding / invoking / and removing multiple EJB instances
from a client. i.e: My test client does the following:
1) Create X number of new instances and store the new
keys in a Vector (get the key from getPrimaryKey())
(Record time it takes for this to occur)
2) For each key in the vector, just perform a ejbFind..()
(Record time it takes for this to occur)
3) For each key in the vector, perform a ejbFind..()
then invoke a method
(Record time it takes for this to occur)
4) For each key in the vector, perform a ejbRemove..()
(Record time it takes for this to occur)
5) Average Time per X
Same simple code works under Jonas and JBoss
In PROD 01, I could create/invoke/remove MANY (> 25000) before an Out of
Memory exception would be thrown (due to a Transaction problem - Marc
emailed and said was fixed) - But all the finds and removes worked perfectly
up to that point.
In PROD 02, I could create/invoke/remove but had the following problem:
When I do a findByPrimaryKey(XXX) of an object I just created, the returned
object isn't aways the one I requested - it would be any one of the other
ones I just created.
Therefore: I would get (from my own code on how I was doing the ejbRemove),
a FinderException or ObjectNotFoundException on the ejbRemove() because the
client would remove the object earlier.
Now... This happens about 80% when the number of instances I am creating is
just 15 instances. I am using BMP & mySQL - which should have no bearing
with the problem.
Same bean works find under PROD 01.
One other thing, the jBoss PROD 2 verifier also states that one
of my deployed bean's Key class needs to override or define the "equals" and
"hashCode" methods - which it does, but I could see
this being a problem, if it can't use the equals ro hashCode right.
Anyone seen this?
Ty Hayden
Ball Aerospace
--
--------------------------------------------------------------
To subscribe: [EMAIL PROTECTED]
To unsubscribe: [EMAIL PROTECTED]
Problems?: [EMAIL PROTECTED]
