So, if
I have an interface MyDao with an implementation class called MyDaoImpl, what
you are saying is that I should never start/commit/end transactions in the
MyDaoImpl class.
But,
if we start referring to iBatis classes from outside the Dao, wouldn't that
defeat the purpose of using a Dao pattern in the first
place?
-----Original Message-----And just to answer another question....
From: Clinton Begin [mailto:[EMAIL PROTECTED]
Sent: Monday, May 09, 2005 4:05 PM
To: [email protected]
Subject: Re: Possible Connection Leak
NEVER call start/commit/end transaction methods from WITHIN your DAO.
Cheers,
Clinton
On 5/9/05, Henry Lu <[EMAIL PROTECTED]> wrote:Exactly.
-Henry
Ming Xue wrote:
>Do you mean I do not need Transaction in SqlMap at all (removing the
>TransactionManager tag) ?
>
>Thanks for any clarification.
>
>-----Original Message-----
>From: Henry Lu [mailto:[EMAIL PROTECTED]]
>Sent: Monday, May 09, 2005 2:23 PM
>To: [email protected]
>Subject: Re: Possible Connection Leak
>
>
>No. You config JNDI in the spring.xml instead.
>
>-Henry
>
>Ming Xue wrote:
>
>
>
>>Hi
>>
>>I am using the SpringDAO framework, basically the SqlMapDAOSupport
>>and SqlMapTemplate, the Transaction is acturally controlled by
>>Spring, in this case, how should I config the transaction of SqlMap,
>>should I use External Type ?
>>
>> <transactionManager type="EXTERNAL">
>> <dataSource type="JNDI">
>> <property name="DataSource" value="java:/PlateauDS"/>
>> </dataSource>
>> </transactionManager>
>>
>>Here is my DAO
>>
>>public class MyDAO extends SqlMapDAOSupport implements MyDAOService
>>
>>{
>> public void myMethod () {
>>
>> getSqlMapClientTemplate().insert("insertAccount", account);
>>
>> }
>>
>>}
>>
>>The transaction is started in the EJB level, which calls the DAO, and
>>a EJB may call any number of DAO to do the job, in this case,
>>do I need to put in the SQLMap.startTransaction in the DAO code ?
>>
>>Thanks
>>Ming
>>
>>
>>
>>
>>[Ming Xue]
>>
>>
>> -----Original Message-----
>>*From:* Clinton Begin [mailto: [EMAIL PROTECTED]]
>>*Sent:* Monday, May 09, 2005 1:04 PM
>>*To:* [email protected]
>>*Subject:* Re: Possible Connection Leak
>>
>> Yes, the approach I suggested should be used regardless of your
>> transaction type.
>>
>> Cheers,
>> Clinton
>>
>> On 5/9/05, *Rafiq, Adnan* <[EMAIL PROTECTED]
>> <mailto:[EMAIL PROTECTED]>> wrote:
>>
>> Thanks Clinton.
>>
>> We are using session beans to manage transactions in our
>> application. Moreover, since we have multiple databases
>> involved we are using XA- drivers.
>>
>> Will the approach you mentioned still work in this scenario?
>>
>> -----Original Message-----
>> *From:* Clinton Begin [mailto:[EMAIL PROTECTED]
>> <mailto:[EMAIL PROTECTED]>]
>> *Sent:* Saturday, May 07, 2005 9:30 PM
>> *To:* [email protected]
>> <mailto:[email protected] >
>> *Subject:* Re: Possible Connection Leak
>>
>>
>> Do ALL of your SqlMap calls follow this pattern:
>>
>> try {
>> sqlMap.startTransaction();
>> //...do work
>> sqlMap.commitTransaction();
>> } finally {
>> sqlMap.endTransaction();
>> }
>>
>> Clinton
>>
>>
>> On 5/6/05, *Rafiq, Adnan* <[EMAIL PROTECTED]
>> <mailto:[EMAIL PROTECTED]>> wrote:
>>
>> I am using a transaction manager type = JTA in my
>> sql-map-config.xml file.
>> After running the application for a while, I am
>> getting the following
>> message in my Weblogic 8.1 console:
>>
>> <May 6, 2005 5:09:08 PM CDT> <Warning> <Common>
>> <BEA-000620> <Forcibly
>> releasing
>> inactive resource
>> "[EMAIL PROTECTED]" back
>> into the pool "TP_BOFA_SYS".>
>>
>> According to BEA, there is a connection leak
>> somewhere. Is it possible that
>> iBatis is not properly closing a connection after
>> obtaining it?
>>
>> Has anyone else come across a similar issue?
>>
>>
>>
>>
>>
>
>
>
>
