Hi
I've got a problem with transaction propagation and pessimistic locking.
The problem is as follows:
A stateless session bean (SB1), with transaction attribute Required, handles
a call sequence.
In that sequence it first calls anoter stateless session bean (SB2) which
has transaction attribute NotSupported.
SB2 gathers some data and performs validation. The data is gathered by
calling en entity bean (EB1), which has
transaction attribute Required. When the calling sequence returns to SB1,
and the data is validated, and it calls
yet another stateless session bean (SB3), which has transaction attribute
RequiresNew.
SB3 now performs some business logic, and then tries to update the data
using EB1 (on the same PK as above).
This results in a timeout (might be deadlock).
The reason of the control flow is that the client call to SB1 should result
in several independent transactions,
each started, and ended in SB3 (which the client can't call).
See call sequence below:
------
Notation:
Bean(Type, Transaction Attribute)
Beans in problem:
SB1(Stateless Session, Required)
SB2(Stateless Session, NotSupported)
SB3(Stateless Session, RequiresNew)
EB1(Entity, Required)
Call sequence:
Seq Caller Operattion Description:
1 Client Calls SB1 Client calls business method
2 SB1 Calls SB2 SB1 Uses SB2 to gather data for validation
3 SB2 Calls EB1 SB2 Uses EB1 to gather data for validation
4 SB1 Calls SB3 SB1 Then calls SB3 to perform the job
5 SB3 Calls EB1 SB3 Tries to use EB1 (Same pk as above) ->
TimeOut, Lock...
6 SB1 Returns to client
------
Isn't this what should happen (concerning the transactions):
(seq above) Transactions
1 Transaction T1 starts
2 Transaction T1 suspends
3 Transaction T2 starts (EB1 gets involved in a transaction)
Transaction T2 ends
4 Transaction T1 resumes
5 Transaction T3 starts (And tries to use EB1)
Transaction T3 ends
6 Transaction T1 ends.
But it looks like transaction T2 doens't start, or that the lock on EB1
doesn't get released,
and therefore EB1 that is used in T2, is still locked.
Is the transaction sequence above correct, or have I misunderstood the
specification?
We are using weblogic, and the documentation for weblogic says:
<Clip from Weblogic Docs.>
The EJB 1.1 container in WebLogic Server Version 5.1 uses a pessimistic
locking mechanism for entity EJB instances. As clients enlist an
EJB or EJB method in a transaction, WebLogic Server places an exclusive
lock on the EJB instance or method for the duration of the transaction.
Other clients requesting the same EJB or method block until the current
transaction completes.
<End clip>
But shouldn't the lock on EB1 be released in seq 3 (according to the EJB
specification)?
Thanks for your help
/Kaj
~~~~~~~~~ ~~~~ ~~~ ~~ ~ ~ ~ ~
[EMAIL PROTECTED]
+46 70 4200148
===========================================================================
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff EJB-INTEREST". For general help, send email to
[EMAIL PROTECTED] and include in the body of the message "help".
