Inside the OJVM this feature is even more important because there is no visual feedback when you are running Java code.
Starting with 10g, I have been using this feature as is explained in my posts Getting logging entering, exiting and throwing work at Oracle JVM, which according to Google Analytics is one of most visited page, but most of tweak requires uploading new classes and configuring it with certain knowledge.
Starting with 12c, the logging feature is directly integrated into the OJVM also for PDB.
This mean that by simply loading a logging.properties file into a user's schema you can control the logging facilities for Java code running with an effective user (Oracle JVM performs specific steps to configure logging options).
For example if I have a class loaded into SCOTT's schema and this class run with an effective user SCOTT, to activate the logging level for that class and user you have to execute:
loadjava -u scott/tiger@mydb -v javavm/lib/logging.propertiesif the logging.properties file includes something like:
$ cat javavm/lib/logging.propertiesremember that for using logging inside the OJVM SCOTT user should have SYS:java.util.logging.LoggingPermission = control, to do that connected as SYS excecute:
# Limit the message that are printed on the console to INFO and above.
java.util.logging.ConsoleHandler.level = ALL
java.util.logging.ConsoleHandler.formatter = java.util.logging.SimpleFormatter
# Scotas OLS
com.scotas.level = INFO
org.apache.level = INFO
$ sqlplus "sys/pwd@mydb as sysdba"also your application can control logging level properties during run time, above configuration are static and defined when the OJVM start the execution of the Java code for a specific DB session.
SQL> exec dbms_java.grant_permission( 'SCOTT', 'SYS:java.util.logging.LoggingPermission', 'control', '');
The example below is Scotas OLS web page available for controlling logging properties during run time (the old logging configuration Servlet of Apache Solr).
Finally where my logs goes?
Independent if you are running your Java application in a traditional RDBMS installation or in a new CDB/PDB configuration log information will goes to the .trc file associated to the session, this is at:
$ORACLE_BASE/diag/rdbms/$ORACLE_SID/$ORACLE_SID/tracethe .trc file is named using this filename convention $ORACLE_SID_jnnn_$PID.trc, for example:
$ cat orclc_j020_11761.trc
Trace file /u01/app/oracle/diag/rdbms/orclc/orclc/trace/orclc_j020_11761.trc
Oracle Database 12c Enterprise Edition Release 184.108.40.206.0 - 64bit Production
With the Partitioning, OLAP, Advanced Analytics and Real Application Testing options
ORACLE_HOME = /u01/app/oracle/product/220.127.116.11.0/dbhome_1
System name: Linux
Node name: localhost
Version: #78-Ubuntu SMP Fri Jul 26 16:21:44 UTC 2013
Instance name: orclc
Redo thread mounted by this instance: 1
Oracle process number: 7
Unix process pid: 11761, image: oracle@pocho (J020)
*** 2013-08-22 11:29:41.805
*** SESSION ID:(387.581) 2013-08-22 11:29:41.805
*** CLIENT ID:() 2013-08-22 11:29:41.805
*** SERVICE NAME:(SYS$USERS) 2013-08-22 11:29:41.805
*** MODULE NAME:(DBMS_SCHEDULER) 2013-08-22 11:29:41.805
*** ACTION NAME:(SOLRSERVLETSERV#1) 2013-08-22 11:29:41.805
*** CONTAINER ID:(3) 2013-08-22 11:29:41.805
*** 2013-08-22 11:52:37.581
INFO: Unset log level on 'root'.
Aug 22, 2013 11:52:37 AM com.scotas.solr.servlet.LogLevelSelection doPost
INFO: Set 'com.scotas' to INFO level.
is important that even if you have multiple PDBs, all log files will go to the trace directory of the CDB because all PDBs re-use the process infra-structure of the CDB.
I will post another blog entry about the impact of the process infra-structure of the Multitenant Environment and the OJVM applications.