Re: Datadirect JDBC driver on DB2: Locks not released
From: Charles Fineman (cfineman-no&spam_at_yahoo.com)
Date: 03/23/04
- Next message: Charles Fineman: "Re: Using nulls in a jdbc Select"
- Previous message: Robert Klemme: "Re: Distributing JDBC Drivers"
- In reply to: Matthew: "Datadirect JDBC driver on DB2: Locks not released"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Tue, 23 Mar 2004 10:08:34 -0500
Are you getting this connection from a managed pool or are you getting
it directly from the driver?
If you are getting it from a pool (i.e. a data source managed by your
app server) then it might be disallowing you from turning auto-commit on
(since it messes with the JTA transaction semantic). All of the
implementations I've encountered silently ignore the setting of
auto-commit (something I invariably test since it's often a source of
errors for developers). I would rather the spec force them to throw an
exception in such cases :-(
Matthew wrote:
> I am involved in a project that is porting a large database
> application from SQL server to DB2 and have run into some difficulty
> with DataDirect's level 3 JDBC driver for DB2 (running under linux).
>
> I have a number of CallableStatements created at the beginning of a
> process, used multiple times during the process and then closed at the
> end (all are created on the same connection - but are used by only one
> thread - i.e. no two statements are executed simultaneously).
>
> One of these statements inserts values into a table and another reads
> from it. The problem is that the statement that inserts acquires a
> lock on the table that isn't released until the statement is CLOSED
> (even though Autocommit is ON). This means that the select statement
> is locked out.
>
> If autocommit is turned off and the connection is commited after the
> insert statement then the locks are released - isn't it true that
> setting autocommit to true should make the connection behave exactly
> as if autocommit was off, but the connection was commited after every
> statement? Is this a bug?
>
> Any help will be appreciated as changing our approach (i.e. manually
> committing) has quite wide impact on the system and is going to cost
> us a lot of development time.
>
> Thanks,
>
> Matthew
> Software Developer
> Mosaic Software
> Cape Town
- Next message: Charles Fineman: "Re: Using nulls in a jdbc Select"
- Previous message: Robert Klemme: "Re: Distributing JDBC Drivers"
- In reply to: Matthew: "Datadirect JDBC driver on DB2: Locks not released"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
