|
Problem |
In your scenario, the client gets a handle to a stateless
session bean, which acts as a facade that calls several stateless and
stateful session beans as a part of the workflow in a transaction aware
environment.
So in this case, should the TimeOut attribute of the session be set larger
than that of the transaction? Also, should the stateless session bean have
a TimeOut set to a much larger value than that of the beans that it calls?
If so, what is the optimal max_value for the timeout?
Is there an algorithm that is needed to set the timeout values in the
preceding scenario? |
|
Solution |
There is no simple algorithm that you can apply. In
general, the bean timeout would be larger than the individual transaction
timeout. In this case, where stateless session bean fronts a group of
other session beans, the correct timeout depends on the relationship
between the beans. If all of the bean instances are created together and
stay together, than all of the beans should have the same timeout.
Therefore, everything goes away at about the same time. If, however, the
contained beans have a lifecycle that is different than the bean that
fronts them, the timeouts should be set appropriately for each individual
bean. |
|
|
|
|
|
|