Weird, why would you need to do that?

Actually you should not need to care about the SQL at all, i mean that's
what the idea of hibernate is all about...

Yes of course, if you set up the mapping accordingly. You would need to set
up a OneToMany/ManyToOne relation with a cascade-all/delete-orphan style of
cascading saves/updates/deletes.

Best regards,
Jan Thomä

On Aug 29, 8:19 am, "sss.z...@gmail.com" <sss.z...@gmail.com> wrote:

That's actually a good question.  We use that all the time at my work
to figure out why the damn thing is broken sometimes.  We're pushing
hibernate to it's limits though.  Too lazy to upgrade I guess.  Moving
on, here's what we do:

In our applicationContext.xml file we have the following hibernate
flags

        <property name="hibernateProperties">
            <props>
                <prop
key="hibernate.dialect">net.sf.hibernate.dialect.OracleDialect</prop>
                <prop key="hibernate.cglib.use_reflection_optimizer">false</prop>
                <!-- ShowSql-->
                <!-- prop key="hibernate.show_sql">true</prop-->
                <!--prop key="hibernate.max_fetch_depth">1</prop-->
            </props>
        </property>

The one you want is SHOW_SQL.  The file name you use may be different
but just set that property and your good.  You will SEE ALL the sql
though.  You'll be amazed.  Oh btw, it's also good to use this
property when you're trying to speed up hibernate and make the queries
more efficient.

-Robert

You specify your own appender for those log messages.

That is just a matter of doing the mapping correct.

Arne

<posted & mailed>

On Thursday 30 August 2007 16:32:36 you wrote:

Okay, you want high performance, high availability. Creating this on your
own is most likely to fail. I am not questioning your coding skills here,
however this is something that is not written by a single person on a
single day. There are free readymade and proven solutions for this, so for
starters I'd go this way:

Put up two linux boxes, install a database on each, maybe mysql. For
availability, you could use vrrpd. Its actually a daemon installed on both
machines, monitoring each other machine. One is master (active) the other
is slave (hot standby). Vrrpd will set up a virtual interface where you can
send requests two. E.g.
two machines: 10.8.0.1 and 10.8.0.2. vrrpd will create a third virtual ip
10.8.0.3. Using this IP you always connect to the current master machine.
In case the master machine fails, it switches, so hot standby becomes
master and master becomes hot standby (which you could repair then). You
should have virtually no interruption.

Next up, set up replication between the two mysql instances.  So all data
written to the master will be replicated into the slave machine.

Now add another two machines where you install your software. Have your
software use the database at the virtual ip address (10.8.0.3). Again
install vrrpd on these two new machines and set it up correctly.

Should be a matter of some hours to read up on the documentation of vrrpd
and mysql and setting it up. As additional benefit, it should work without
any special modifications needed to be coded into your program.

Just my $0.02..

Best regards,
Jan