Oracle Package Initialization

Did you notice that Oracle PL/SQL packages can have an initialization section? The initialization part of a package is run only once, the first time you reference the package. Here is an example:

CREATE PACKAGE emp_actions AS
     /* Declare externally callable subprograms. */
     FUNCTION hire_employee (
        ename  VARCHAR2,
        job    VARCHAR2,
        mgr    REAL,
        sal    REAL,
        comm   REAL,
        deptno REAL) RETURN INT;
  END emp_actions;

  CREATE PACKAGE BODY emp_actions AS
     number_hired INT;  -- visible only in this package

     /* Fully define subprograms specified in package. */
     FUNCTION hire_employee (
        ename  VARCHAR2,
        job    VARCHAR2,
        mgr    REAL,
        sal    REAL,
        comm   REAL,
        deptno REAL) RETURN INT IS
        new_empno INT;
     BEGIN
        SELECT empno_seq.NEXTVAL INTO new_empno
        FROM dual;
        INSERT INTO emp VALUES (new_empno, ename, job,
           mgr, SYSDATE, sal, comm, deptno);
        number_hired := number_hired + 1;
        RETURN new_empno;
     END hire_employee;

  BEGIN  -- initialization part starts here
     INSERT INTO emp_audit
     VALUES (SYSDATE, USER, 'EMP_ACTIONS');
     number_hired := 0;
  END emp_actions;

The initialization part of a package is run just once, the first time you reference the package. So, in the last example, only one row is inserted into the database table emp_audit. Likewise, the variable number_hired is initialized only once.

Every time the procedure hire_employee is called, the variable number_hired is updated. However, the count kept by number_hired is session specific. That is, the count reflects the number of new employees processed by one user, not the number processed by all users.

Using rsync to Copy Oracle Home Directories

If you're using UNIX for your ORACLE database server, it is possible to copy the Oracle Database Software (the Oracle Home directory) from one server to another using rsync.

Here is the example rsync syntax to copy the Oracle Home from one server (svr1) to another server (svr2). This process will ensure that all symbolic links and files owned by root get copied over correctly. Login to the source database server (svr1) as the root user account and run the following:

[root@svr1 ~]# rsync -auvzpogl -e ssh /u01/app/oracle/product/10.2.0/db_1/ svr2:/u01/app/oracle/product/10.2.0/db_1/

Sending E-Mail in PL/SQL

Here some example code you can use for sending email from within PL/SQL

CREATE OR REPLACE PROCEDURE SEND_MAIL_TCP (
    msg_from     VARCHAR2 := 'sender@testing.com'
  , msg_to       VARCHAR
  , msg_subject  VARCHAR2 := 'E-Mail message from your database'
  , msg_text     VARCHAR2 := ''
)
IS
  c   UTL_TCP.CONNECTION;
  rc  INTEGER;
BEGIN
  c  := UTL_TCP.OPEN_CONNECTION('localhost', 25);       -- open the SMTP port 25 on local machine
  rc := UTL_TCP.WRITE_LINE(c, 'HELO localhost');
  rc := UTL_TCP.WRITE_LINE(c, 'MAIL FROM: '||msg_from);
  rc := UTL_TCP.WRITE_LINE(c, 'RCPT TO: '||msg_to);
  rc := UTL_TCP.WRITE_LINE(c, 'DATA');                  -- Start message body
  rc := UTL_TCP.WRITE_LINE(c, 'Subject: '||msg_subject);
  rc := UTL_TCP.WRITE_LINE(c, '');
  rc := UTL_TCP.WRITE_LINE(c, msg_text);
  rc := UTL_TCP.WRITE_LINE(c, '.');                     -- End of message body
  rc := UTL_TCP.WRITE_LINE(c, 'QUIT');
  UTL_TCP.CLOSE_CONNECTION(c);                          -- Close the connection
EXCEPTION
  WHEN others THEN
    RAISE_APPLICATION_ERROR(-20000, 'Unable to send e-mail message from PL/SQL routine.');
END;

Another Oracle apps DBA Interview Questions (with answer)

Questions:

1. What happens if the ICM goes down?
2. How will you speed up the patching process?
3. How will you handle an error during patching?
4. Provide a high-level overview of the cloning process and post-clone manual steps.
5. Provide an introduction to AutoConfig. How does AutoConfig know which value from the XML file needs to be put in which file?
6. Can you tell me a few tests you will do to troubleshoot self-service login problems? Which profile options and files will you check?
7. What could be wrong if you are unable to view concurrent manager log and output files?
8. How will you change the location of concurrent manager log and output files?
9. If the user is experiencing performance issues, how will you go about finding the cause?
10. How will you change the apps password?
11. Provide the location of the DBC file and explain its significance and how applications know the name of the DBC file.

Answers:

1. All the other managers will keep working. ICM only takes care of the queue control requests, which means starting up and shutting down other concurrent managers.
2. 
   
   • You can merge multiple patches.
   • You can create a response file for non-interactive patching.
   • You can apply patches with options (nocompiledb, nomaintainmrc, nocompilejsp) and run these once after applying all the patches.
3. Look at the log of the failed worker, identify and rectify the error and restart the worker using adctrl utility.
4. Run pre-clone on the source (all tiers), duplicate the DB using RMAN (or restore the DB from a hot or cold backup), copy the file systems and then run post-clone on the target (all tiers).
   Manual steps (there can be many more):
   
   • Change all non-site profile option values (RapidClone only changes site-level profile options).
   • Modify workflow and concurrent manager tables.
   • Change printers.
5. AutoConfig uses a context file to maintain key configuration files. A context file is an XML file in the $APPL_TOP/admin directory and is the centralized repository.
   When you run AutoConfig it reads the XML files and creates all the AutoConfig managed configuration files.
   For each configuration file maintained by AutoConfig, there exists a template file which determines which values to pick from the XML file.
6. 
   
   • Check guest user/password in the DBC file, profile option guest user/password, the DB.
   • Check whether apache/jserv is up.
   • Run IsItWorking, FND_WEB.PING, aoljtest, etc.
7. Most likely the FNDFS listener is down. Look at the value of OUTFILE_NODE_NAME and LOGFILE_NODE_NAME in the FND_CONCURRENT_REQUESTS table. Look at the FND_NODES table. Look at the FNDFS_ entry in tnsnames.ora.
8. The location of log files is determined by parameter $APPLCSF/$APPLLOG and that of output files by $APPLCSF/$APPLOUT.
9. 
   
   • Trace his session (with waits) and use tkprof to analyze the trace file.
   • Take a statspack report and analyze it.
   • O/s monitoring using top/iostat/sar/vmstat.
   • Check for any network bottleneck by using basic tests like ping results.
10. 
    
    • Use FNDCPASS to change APPS password.
    • Manually modify wdbsvr.app/cgiCMD.dat files.
    • Change any DB links pointing from other instances.
11. 
    
    • Location: $FND_TOP/secure directory.
    • Significance: Points to the DB server amongst other things.
    • The application knows the name of the DBC file by using profile option "Applications Database Id."

Five DBA Best Practices

Multiple Oracle Homes
My favorite best practice is the one about multiple Oracle Homes. Here it how it goes. When applying a patch or a patchset, I recommend against applying to the existing Oracle Home. Instead, I suggest creating a new Oracle Home, and apply the patches there.

I create the first Oracle Home at /app/oracle/db_1, for instance. When a patch comes out, I install the whole Oracle software in a different home -- /app/oracle/db_2 -- and then apply the patch there. During the process of installation and patch application, the database is up and running, as it runs off the home /db_1. When the outage window comes, all I have to do is to shut down Oracle; change Oracle Home to db_2 and bring the database up. If there is a problem, I can reset the Oracle Home back to the old one.

So, here is the conventional approach:

1. Shut down the database
2. Apply patch to the Oracle Home
3. Start the database
4. In case of problems:
5. Shut down the database
6. Roll back the patch
7. Start the database

Steps 2 and 6 could take as much as three hours depending on the amount of patching. The database is down during these times.

In the new approach:

1. Install new Oracle Home
2. Apply the patch to the new Home
3. Shut down the database
4. Change Oracle Home to the new location
5. Start the database
6. In case of problems:
7. Shut down the database
8. Change Oracle Home to the old one
9. Start the database

The database is down only during steps 4 and 8, which takes a couple of minutes at the most, not hours.

So, here are the advantages:

1. The downtime is significantly reduced, to one 60th of the original time.
2. The risk is reduced significantly, as there is no rolling back a patch; you just go to the older version.
3. You can perform a "diff" on these two homes to see what changed. You can see the differences across multiple homes as well.
4. You can take several databases running on the same server to the new Oracle Home one by one.
5. You can see the various Oracle Homes and what patch level they are on using the inventory.

The only negative part is the space consumption -- you need space for two Oracle Homes. But considering a typical Oracle Home takes about 4 GB or less, this aspect of the suggestion is trivial.

Set audit trail to DB
Set the audit trail to DB by placing the parameter AUDIT_TRAIL = DB in the initialization parameter file during the database creation. Setting this parameter does not start the auditing, because an explicit AUDIT command must be given on the object. But the parameter must be set to a value other than FALSE (the default) for the command to take effect. Being a non-dynamic parameter, the database must be bounced to change the value of AUDIT_TRAIL. To save the trouble and avoid an outage, always set the value to DB, even if you never intend to audit anything. It does not break anything and you will always be ready to audit when the time comes.

Don't use .log
Don't use .log as the extension of redo logs. Someone may run a script to remove all the log files assuming they are redundant and you will end up losing the online redo logs as well, forcing a database recovery. Instead, name them with extension "redo" or "rdo."

Preview RMAN Restore
Preview RMAN Restore to identify all the various backup pieces that will be used in the recovery process without doing an actual recovery. This eliminates any surprises from missing pieces during an actual recovery process.

Create a new Oracle user for clients running on the same server as the DB
The Oracle Database server software also contains the client piece, which allows the clients to connect to the database on the same server. But as a best practice do not use the same user or the software; use a new one. For instance, if "oracle" is the user to install Oracle software, create a new user called, say, "oraapp" and install the client-only software using that user. The user "oraapp" should not be part of the dba or the oinstall group; so this user can't log on to the database as sysdba. Create a new group called "appgrp" and assign the user oraaap to this group. All the application users on the box should also be part of the appgrp group. This way they can use the sqlplus, sqlldr and other executables on the server, but be able to connect as sysdba.

The common practice is to use the client software in the same user as the database software owner; but starting with 10.2, Oracle has changed the security policy that takes away the global execution permissions from the Oracle Home. So the only option is to let app users be part of the dba group or change the permissions on Oracle Home -- both make the database extremely vulnerable.

dbms_backup_restore Package

The dbms_backup_restore package is used as a PL/SQL command-line interface for replacing native RMAN commands, and it has very little documentation.

The Oracle docs note how to install and configure the dbms_backup_restore package:

“The DBMS_BACKUP_RESTORE package is an internal package created by the dbmsbkrs.sql and prvtbkrs.plb scripts. This package, along with the target database version of DBMS_RCVMAN, is automatically installed in every Oracle database when the catproc.sql script is run. This package interfaces with the Oracle database server and the operating system to provide the I/O services for backup and restore operations as directed by RMAN.”

The docs also note that “The DBMS_BACKUP_RESTORE package has a PL/SQL procedure to normalize filenames on Windows NT platforms.”

Oracle DBA John Parker gives this example of dbms_backup_restore to recover a controlfile:

declare
devtype varchar2(256);
done boolean;
begin
devtype:=dbms_backup_restore.deviceallocate( type=>'sbt_tape',
params=>'ENV=(OB2BARTYPE=Oracle8,OB2APPNAME=rdcs,OB2BARLIST=ORA_RDCS_WEEKLY)',
ident=>'t1');
dbms_backup_restore.restoresetdatafile;
dbms_backup_restore.restorecontrolfileto('D:\oracle\ora81\dbs\CTL1rdcs.ORA');
dbms_backup_restore.restorebackuppiece(
'ORA_RDCS_WEEKLY.dbf', DONE=>done );
dbms_backup_restore.restoresetdatafile;
dbms_backup_restore.restorecontrolfileto('D:\DBS\RDCS\CTL2RDCS.ORA');
dbms_backup_restore.restorebackuppiece(
'ORA_RDCS_WEEKLY.dbf', DONE=>done );
dbms_backup_restore.devicedeallocate('t1');
end;

Here are some other examples of using dbms_backup_restore:

DECLARE
devtype varchar2(256);
done boolean;
BEGIN
devtype := dbms_backup_restore.DeviceAllocate (type => '',ident => 'FUN');
dbms_backup_restore.RestoreSetDatafile;
dbms_backup_restore.RestoreDatafileTo(dfnumber => 1,toname => 'D:\ORACLE_BASE\datafiles\SYSTEM01.DBF');
dbms_backup_restore.RestoreDatafileTo(dfnumber => 2,toname => 'D:\ORACLE_BASE\datafiles\UNDOTBS.DBF');
--dbms_backup_restore.RestoreDatafileTo(dfnumber => 3,toname => 'D:\ORACLE_BASE\datafiles\MYSPACE.DBF');
dbms_backup_restore.RestoreBackupPiece(done => done,handle => 'D:\ORACLE_BASE\RMAN_BACKUP\MYDB_DF_BCK05H2LLQP_1_1', params => null);
dbms_backup_restore.DeviceDeallocate;
END;
/

--restore archived redolog
DECLARE
devtype varchar2(256);
done boolean;
BEGIN
devtype := dbms_backup_restore.DeviceAllocate (type => '',ident => 'FUN');
dbms_backup_restore.RestoreSetArchivedLog(destination=>'D:\ORACLE_BASE\achive\');
dbms_backup_restore.RestoreArchivedLog(thread=>1,sequence=>1);
dbms_backup_restore.RestoreArchivedLog(thread=>1,sequence=>2);
dbms_backup_restore.RestoreArchivedLog(thread=>1,sequence=>3);
dbms_backup_restore.RestoreBackupPiece(done => done,handle => 'D:\ORACLE_BASE\RMAN_BACKUP\MYDB_LOG_BCK0DH1JGND_1_1', params => null);
dbms_backup_restore.DeviceDeallocate;
END;
/