Re: [OJB-developers] ERROR: Server configuration denies access to data source

[<ri...@ya...>]

Hello,

It turns out that this error:
  java.sql.SQLException: Server configuration denies access to data =
source

was caused because MySql used its "max_connections" limit of 100.

I tried to initialize MySql with a parameter of max_connections=3D1000, =
but it gave me the same error.

What leads me to think that it's a problem with OJB not closing =
connections.

It looks like OJB open a lot of connections trying to materialize the =
dynamic proxies, but then it stops and gives the Exception, because all =
the connections are occupied.

Why does OJB open so many connections when using dynamic proxies?

Thanks in advance,

Ricardo Arguello


----- Original Message -----=20
From: "Mahler Thomas" <tho...@it...>
To: "'Ricardo Arg=FCello'" <ri...@ya...>
Cc: "Objectbridge (E-Mail)" =
<obj...@li...>
Sent: Tuesday, February 19, 2002 4:48 AM
Subject: AW: [OJB-developers] ERROR: Server configuration denies access =
to data source


> Hi Ricardo,
>=20
> >=20
> > Hello,
> >=20
> > Since OJB is performing REALLY slowly, I tried to add the=20
> > next tag to my repository.xml file:
> >=20
> >   <class.proxy>dynamic</class.proxy>
> >=20
> > in every <ClassDescriptor> tag.
> >=20
> > All my objects implement an interface, so this theorically=20
> > should work fine.
> > That's the only thing I changed in this file, and now I'm=20
> > getting this strange exception:
> =20
> > [DEFAULT] ERROR: Server configuration denies access to data source
> > java.sql.SQLException: Server configuration denies access to=20
> > data source
> >         at org.gjt.mm.mysql.MysqlIO.init(Unknown Source)
> >         at org.gjt.mm.mysql.Connection.connectionInit(Unknown =
Source)
> >         at=20
> > org.gjt.mm.mysql.jdbc2.Connection.connectionInit(Unknown Source)
> >         at org.gjt.mm.mysql.Driver.connect(Unknown Source)
> >         at=20
> > java.sql.DriverManager.getConnection(DriverManager.java:517)
> >         at=20
> > java.sql.DriverManager.getConnection(DriverManager.java:177)
> >         at=20
> > =
ojb.broker.accesslayer.ConnectionFactoryDefaultImpl.newConnection(Con
> > nectionFactoryDefaultImpl.java:72)
> <snip>
>=20
> OK this exception is raised during the materilization of a proxy. But =
the
> error is NOT Proxy related at all. You get an Error from the the MySQL
> driver.
> I'm not a MySQL expert thus I can't help here.
>=20
> >=20
> >=20
> > --------------------------------------------------------------
> > ------------
> >=20
> > After a LOT of these messages: =20
> >=20
> > [DEFAULT] WARN: problems with platform=20
> > ojb.broker.platforms.PlatformMySQLImpl:
> > ojb.broker.platforms.PlatformMySQLImpl
> > [DEFAULT] WARN: OJB will use PlatformDefaultImpl instead
> >=20
>=20
> This is totally harmless. It just tells you that OJB does not use any =
MySql
> specific PlatForm code.
>=20
> >=20
> >=20
> > Do you have any idea of what's might the problem be?
>=20
> IMHO this is a MySQL specific problem.
> =20
> > Besides this, how can I make OJB perform faster?
> >=20
>=20
> Mhh.
> Of course using an O/R layer has a certain impact on performance.
> If you are using handcoded JDBC and handcoded mappings you can do =
things
> much faster.
> There is nothing like a free lunch !
>=20
> But: has been carefully designed to minimize the performance impact.
> There is a special JUnit target 'performance'
>=20
> By calling 'build.sh performance' you can execute the OJB performance
> testsuite.
> The testsuite runs to tests:
> 1. running five tescenarios using OJB
> 2. running the same scenarios with manual coded JDBC calls.
>=20
> If you are executing it out of the box the tests are executed against =
the
> HSQLDB database that ships with OJB.
> You will see that the OJB mode is about half as fast as the native =
JDBC
> mode.
> (For instance inserting 10.000 Objects takes about 2500 msecs using =
JDBC and
> about 5000 msecs using OJB on one of my machines.)
>=20
> This sounds bad. But you have to take in account that HDSQL does =
everything
> in memory and that there is no network overhead involved.
>=20
> Now just change the target database (in the repository.xml) to say MS =
access
> and you will see a totally different picture:
> Inserting 10.000 object via JDBC takes 50.000 msecs and 52.500 msecs =
using
> OJB.
>=20
> You see: the absolute OJB overhead remains the same but the factor =
OJB/JDBC
> decreases enormously.
>=20
> Please try this performance test against your target database and tell =
us
> your exact results. I am almost sure that your results will look quite
> similar!
>=20
>=20
> On the other hand: There has been a lot of work on OJB internals in =
the last
> months and it may well be that these refactorings had a negative =
performance
> impact. This must be carefully checked!
> We will need another performance tuning code-walkthrough before we can
> release OJB 1.0.
>=20
> HTH,
>=20
> Thomas
>=20
> >=20
> > Thanks in advance!
> >=20
> > Ricardo Arguello
> >=20
> >=20
> > _______________________________________________
> > Objectbridge-developers mailing list
> > Obj...@li...
> > https://lists.sourceforge.net/lists/listinfo/objectbridge-developers
> >=20
>