Database connection pool in Java

[Erik Pragt]

Hi all,

I was wondering, what's the correct way of using connections to an OrientDB from my Java webapp? According to the OrientDB book, it's to use ODatabaseDocumentPool.global(), which is deprectated. Currently, I just create a new connection every time, something like this:

public void setUserName(String name) {

  ODatabaseDocumentTx db = new ODatabaseDocumentTx("remote:localhost/demo").open("demo", "demo")
  db.command(new OCommandSQL("update User set name=?")).execute(name);
  db.close();

}

But this seems hardly the best way. What's the idiomatic way of handling connections in OrientDB?

Thanks,

Erik

[Erik Pragt]

Btw, even the documentation, as listed here: http://orientdb.com/docs/2.0/orientdb.wiki/Document-Database.html, still recommends using the deprecated  ODatabaseDocumentPool.global() code. Is this correct?

[machak]

Hi Eric,

On Saturday, December 26, 2015 at 7:58:38 PM UTC+1, Erik Pragt wrote:

Btw, even the documentation, as listed here: http://orientdb.com/docs/2.0/orientdb.wiki/Document-Database.html, still recommends using the deprecated  ODatabaseDocumentPool.global() code. Is this correct?

you could use:

pool = new OPartitionedDatabasePool(getUrl(), getUsername(), getPassword(), getMaxPoolSize());

public ODatabaseDocumentTx openDatabase() {
    return pool.acquire();
}

 

cheers

/m

[Erik Pragt]

Hi /m,

Yes, I think that would work, thanks! And when I'm done with the connection, can I just close it (or use an ARM block) ?

Thanks, Erik

[machak]

On Saturday, December 26, 2015 at 11:25:10 PM UTC+1, Erik Pragt wrote:

Hi /m,

Yes, I think that would work, thanks! And when I'm done with the connection, can I just close it (or use an ARM block) ?

yes, I tend to use finally block to close it (dunno if it implements closeable interface stuff for auto resources...probably it does)

cheers

/m 

[Erik Pragt]

Okay, thanks! (And yes, it implements closeable, else it cannot be used in an ARM block....)

[Erik Pragt]

Hi /m,

I tested this, and it doesn't work :-(

I was using this: 

OObjectDatabaseTx acquire = OObjectDatabasePool.global().acquire();, 

but it's deprected. Now I have to use this:

OPartitionedDatabasePool oPartitionedDatabasePool = new OPartitionedDatabasePool("x", "x", "y");
ODatabaseDocumentTx acquire1 = oPartitionedDatabasePool.acquire();

But that's returning a ODatabaseDocumentTx, while I need an OObjectDatabaseTx.

The only workaround I've found so far, is to use this:

OPartitionedDatabasePool oPartitionedDatabasePool = new OPartitionedDatabasePool("x", "x", "y");
OObjectDatabaseTx acquire1 = new OObjectDatabaseTx(oPartitionedDatabasePool.acquire());

But it looks like a pretty expensive operation, if I look at the sourcecode. Is this the way to go?

Thanks,

Erik

On Saturday, December 26, 2015 at 10:22:05 PM UTC+1, machak wrote:

[social...@gmail.com]

Hi Erik,

I am not expert but i believe it is not right way to use.

OObjectDatabaseTx : it should be use for object API. 
for Graph API you should use : OrientGraphFactory, OrientGraph

Thanks 

[Erik Pragt]

Hi,

Thanks for your reply. I know the OObjectDatabaseTx is used for the Object API, that's what I'm using, but that doesn't my question on how to use the connection pool for the object API.

Cheers, Erik

[social...@gmail.com]

try this :
OObjectDatabasePool pool = new OObjectDatabasePool(getUrl(), getUsername(), getPassword());
        pool.setup(minPoolSize, maxPoolSize);
OObjectDatabaseTx db = pool.acquire();

OObjectDatabasePool is Deprecated
thanks

[machak]

On Sunday, December 27, 2015 at 1:55:55 AM UTC+1, Erik Pragt wrote:

Hi /m,

I tested this, and it doesn't work :-(

I was using this: 

OObjectDatabaseTx acquire = OObjectDatabasePool.global().acquire();, 

but it's deprected. Now I have to use this:

OPartitionedDatabasePool oPartitionedDatabasePool = new OPartitionedDatabasePool("x", "x", "y");
ODatabaseDocumentTx acquire1 = oPartitionedDatabasePool.acquire();

But that's returning a ODatabaseDocumentTx, while I need an OObjectDatabaseTx.

The only workaround I've found so far, is to use this:

OPartitionedDatabasePool oPartitionedDatabasePool = new OPartitionedDatabasePool("x", "x", "y");
OObjectDatabaseTx acquire1 = new OObjectDatabaseTx(oPartitionedDatabasePool.acquire());

But it looks like a pretty expensive operation, if I look at the sourcecode. Is this the way to go?

I think this is correct usage, at least looking into implementation part of it. OObjectDatabaseTx is documented as a wrapper class and does some initializing (entity manager & object serializer helpers). 

I believe overhead is not that big. 

cheers,

/m

[Erik Pragt]

This doesn't even compile.

[machak]

On Sunday, December 27, 2015 at 7:16:16 PM UTC+1, machak wrote:

On Sunday, December 27, 2015 at 1:55:55 AM UTC+1, Erik Pragt wrote:

Hi /m,

I tested this, and it doesn't work :-(

I was using this: 

OObjectDatabaseTx acquire = OObjectDatabasePool.global().acquire();, 

but it's deprected. Now I have to use this:

OPartitionedDatabasePool oPartitionedDatabasePool = new OPartitionedDatabasePool("x", "x", "y");
ODatabaseDocumentTx acquire1 = oPartitionedDatabasePool.acquire();

But that's returning a ODatabaseDocumentTx, while I need an OObjectDatabaseTx.

The only workaround I've found so far, is to use this:

OPartitionedDatabasePool oPartitionedDatabasePool = new OPartitionedDatabasePool("x", "x", "y");
OObjectDatabaseTx acquire1 = new OObjectDatabaseTx(oPartitionedDatabasePool.acquire());

But it looks like a pretty expensive operation, if I look at the sourcecode. Is this the way to go?

I think this is correct usage, at least looking into implementation part of it. OObjectDatabaseTx is documented as a wrapper class and does some initializing (entity manager & object serializer helpers). 

I believe overhead is not that big. 

cheers,

just wanted to add: I am using orientdb for a while now 2+ years I think, but only graph part and I have to say API is often confusing and could be simplified hier and there. 

e.g. by providing clear naming/classes for each DB pool would help a lot....something like  new ObjectDatabasePool() or new GraphDatabasePool()....new DocumentdatabasePool() would be a good and clear abstraction and you wouldn't need to figure out what Partitioned means in OPartitionedDatabasePool....especially there is no "ODatabasePool",...go figure...also: 

part around pooled connections changed (and implementation improved) a lot in recent versions, however there is still a lot of legacy, deprecated stuff lingering around and makes things even more confusing / difficult. 

[Erik Pragt]

On Sunday, December 27, 2015 at 7:16:16 PM UTC+1, machak wrote:

On Sunday, December 27, 2015 at 1:55:55 AM UTC+1, Erik Pragt wrote:

Hi /m,

I tested this, and it doesn't work :-(

I was using this: 

OObjectDatabaseTx acquire = OObjectDatabasePool.global().acquire();, 

but it's deprected. Now I have to use this:

OPartitionedDatabasePool oPartitionedDatabasePool = new OPartitionedDatabasePool("x", "x", "y");
ODatabaseDocumentTx acquire1 = oPartitionedDatabasePool.acquire();

But that's returning a ODatabaseDocumentTx, while I need an OObjectDatabaseTx.

The only workaround I've found so far, is to use this:

OPartitionedDatabasePool oPartitionedDatabasePool = new OPartitionedDatabasePool("x", "x", "y");
OObjectDatabaseTx acquire1 = new OObjectDatabaseTx(oPartitionedDatabasePool.acquire());

But it looks like a pretty expensive operation, if I look at the sourcecode. Is this the way to go?

I think this is correct usage, at least looking into implementation part of it. OObjectDatabaseTx is documented as a wrapper class and does some initializing (entity manager & object serializer helpers). 

I believe overhead is not that big. 

cheers,

/m

Well, I didn't measure it, but when I look at the init method of the OObjectDatabaseTxt, it doesn't look like nothing, but until I've found something better, I'll keep it like this. Thanks for the help!

Erik