Multithreaded connections running sql against cartridge

Hi

I'm having a bit of a problem running multiple threads (each with their own jdbc connection) deduping mols against a cartridge DB. It works on several of our databases, but one DB in particular throws the following error for any threads above 1 (first thread works, all the others fail).

java.sql.SQLException: ORA-29902: error in executing ODCIIndexStart() routine

ORA-29532: Java call terminated by uncaught Java exception: java.rmi.ServerException: RemoteException occurred in server thread; nested exception is: 

	java.rmi.RemoteException: java.sql.SQLException: ORA-01017: invalid username/password; logon denied

ORA-06512: at "JCHEM.JCHEM_CORE_PKG", line 89

ORA-06512: at "JCHEM.JC_IDXTYPE_IM", line 298

 

As far as our DBA can tell, all the machines are setup identically.

Oracle environment:

Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - 64bit Production

PL/SQL Release 11.2.0.1.0 - Production

CORE    11.2.0.1.0      Production

TNS for Linux: Version 11.2.0.1.0 - Production

NLSRTL Version 11.2.0.1.0 - Production



JChem owner: JCHEM



JChem Server environment:

Java VM vendor: Sun Microsystems Inc.



JCHEM_CORE_PKG.GETENVIRONMENT()

--------------------------------------------------------------------------------

Java version: 1.6.0_18

Java VM version: 16.0-b13

JChem version: 5.12.4

JChem Index version: 5120000

JDBC driver version: 11.1.0.7.0-Production

 

Any ideas?

Thanks!

 

Hi,

What do you call a "DB"? An Oracle instance? A JChem Cartridge instance?

Is this happening on an Oracle Real Application cluster?

Thank you,

Peter

Hi

We have 3 oracle instances (dev/test/prod) all running on their own linux box and on each there is a jchem cartridge installed. 

The above problem only seems to happen on our prod machine. Dev and test work fine (although they are all suppose to be identical).

 

Also, if i step through my code for the first few sql queries, it seems to prevent the errors, but then the queries do not run in parallel. ie. If I have a 1k mol file, and i step through the first 20 mols (jc_compare with smiles), and then continue to run the app as normal, there are no error messages, but it runs at the same speed as 1 thread (a single connection).  On the other machines (dev/test), there is a speed up with multiple threads. 

 

Thanks.

Would the jchem_core_pkg.use_password procedure not be thread-safe? Hmm looks like its entry point is synchronized... Do you use the jchem_core_pkg.use_password procedure at the beginning of each session, correct?

Peter

use_password solved the problem, although I didn't call use_password on any of the other databases.

 

Do you have any guesses about what settings would cause use_password to have to be called?

Another application using JChem Cartridge, perhaps?

use_password needs (by default) to be called only once per user during the life-time of the JChem Cartridge server -- another application may have called it before yours.

Peter