This assumes that you have ONLY opened port 8000 for APPS login access.

For any other access I can only recommend to use VPN as a full OEL and RDBMS lock down is very complex.

I used the following script (use at your own risk):

begin
    /* lock ALL seeded accounts first */
    update fnd_user set end_date=sysdate;
    commit;
    /* then unlock the necessary ones */
    update fnd_user set end_date=null where user_name in (
        ‘ANONYMOUS’,
        ‘APPSMGR’,
        ‘AUTOINSTALL’,
        ‘CONCURRENT MANAGER’,
        ‘FEEDER SYSTEM’,
        ‘GUEST’, /* NEVER touch this account as this enables APPS login */
        ‘INITIAL SETUP’,
        ‘ORACLE’, /* my demo account */ 
        ‘SYSADMIN’
    );
    commit;
    /* set password – but NOT for GUEST */
    fnd_user_pkg.updateuser(‘ANONYMOUS’,null,’password’);
    fnd_user_pkg.updateuser(‘APPSMGR’,null,’password’);
    fnd_user_pkg.updateuser(‘AUTOINSTALL’,null,’password’);
    fnd_user_pkg.updateuser(‘CONCURRENT MANAGER’,null,’password’);
    fnd_user_pkg.updateuser(‘FEEDER SYSTEM’,null,’password’);
    fnd_user_pkg.updateuser(‘INITIAL SETUP’,null,’password’);
    fnd_user_pkg.updateuser(‘SYSADMIN’,null,’password’);
    fnd_user_pkg.updateuser(‘ORACLE’,null,’password’);
    commit;
end;
/
exit;

Note: Do NOT to touch the GUEST user.

No problem – pick up the R12 installation described earlier and install the hard disk in a HTPC!

For the purpose I bought a Acer AspireRevo R3700 with the following relevant specifications:

• Processor Intel® Atom D525 processor 1.8GHz
• Memory 2GB DDR3 800MHz
• Graphics NVIDIA ION graphics solution
• Networking Gigabit Ethernet, Wake-on-LAN ready and WLAN: 802.11b/g/n

Replace the built-in harddisk with the R12 harddisk and boot the machine.

The machine can be accessed from your network using Putty or Xming.

The only problem experienced was to fix the Realtek LAN networking.

A Linux driver is needed for the network card Realtek R8168.

Unpack the file and follow the instructions in the readme file.

Don’t forget to open port 8000 in your firewall and lock down your FND USER passwords.

Conclusion

Performance is good and rather improved thanks to the internal SATA connection which is much faster than using an external USB 2.0 connection.

The dual core Intel Atom D525 netbook CPU seems unimpressed with the R12 technical stack and performs well.

Running one Java client and a concurrent manager report does not stress the CPU.

The R12 system with a single user logged on tops out around 1.7Gb RAM so the built-in 2Gb RAM is plenty.

Note the hyper threading shows as additional CPU’s:

In 11i we were securing transactions by Operating Unit (OU) only and we would have to choose a corresponding responsibility to see the transactions:

• OU1 – Receivables Manager
• OU2 – Receivables Manager
• OU3 – Receivables Manager

In R12 it is now possible to have a shared services responsibility like:

• ALL – Receivables Manager

This new kind of security is based on Security Profiles which is a hierarchy of operating units.

So best practice in R12 is now to secure all responsibilities by Security Profile rather than by Operating Unit.

11i non-MOAC setup:

So a global shared service responsibility was not possible in 11i so some companies had very long lists of responsibilities.

Also note in R11i site level setting takes precedence at lower levels so the responsibility level setting in this case is not needed – only if the OU had been different.

R12 MOAC setup:

Assuming we have a hierarchy like this where OU and security profiles have same names:

> Vision Global
>> Vision US
>>> Vision Operations (OU)
>>> Vision Services (OU)
>> Vision Europe (OU)

We do not need to have a security profile for all branches in the hierarchy above but it is best practice to do so.

So in R12 GL Set of Books Name is gone and you leave MO: Operating Unit empty and leave it to the Security Profile.

Note: MO: Default Operating Unit is not required and can be left empty however I have had several bugs in R12.1.1 and R12.1.2 related to missing default operating unit in IEX so try leaving it empty first and if you get problems put a default value in this and try again.

So Oracle has created MOAC but forgot a few details here and there.

Mixed R11i/R12 MOAC setup:

Some modules in R12 does not yet fully support MOAC so a mixed setup may be needed.

Luckily Oracle changed the MO profile settings inheritance so the site level setting does NOT override at responsibility level as in R11i – well at least this setup worked in R12.1.1.

Using the setup above you will get a site level default OU of “Vision Operations” however at responsibility level the security profile takes precedence so Receivables (Global) will be able to see all OU’s within “Vision Global”. In 11i you would have been locked down to “Vision Operations”.

Technical

11i – View Based Security

11i used views to secure transactions in individual schemas:

APPS.AR_PAYMENT_SCHEDULES on AR.AR_PAYMENT_SCHEDULES_ALL

With where clause using CLIENT_INFO set from the responsibility context.

In PL/SQL programs you would set:

• fnd_global.apps_initialize( var_user_id, var_resp_id, var_application_id);

and:

• dbms_application_info.set_client_info(var_organization_id);

R12 – VPD Based Security

R12 uses the Virtual Private Database (VPD) technology for MOAC security.

Synonym APPS.AR_PAYMENT_SCHEDULES on AR.AR_PAYMENT_SCHEDULES_ALL

With a policy attached to synonym APPS.AR_PAYMENT_SCHEDULES.

In PL/SQL program you still need to set:

• fnd_global.apps_initialize( var_user_id, var_resp_id, var_application_id);

and for single OU access:

• mo_global.set_policy_context(‘S’,var_organization_id);

or for multi OU access:

• mo_global.set_policy_context(‘M’,null);

Accessible Organisation Units

In 11i it was easy to see what access you had:

• dbms_application_info.read_client_info(var_organization_id);

Which would return your current org_id however in R12 you need to do the following:

• select organization_id
  from HR_ORGANIZATION_UNITS
  where MO_GLOBAL.CHECK_ACCESS(organization_id)=’Y’
  ;

I usually use the following script:

declare

  l_user_id number;

  l_responsibility_id number;

  l_application_id number;

  l_organization_id number;

begin

--  dbms_output.enable;

  select application_id,responsibility_id

    into l_application_id,l_responsibility_id

    from fnd_responsibility_tl

    where responsibility_name = 'General Ledger Super User';

  select user_id

    into l_user_id

    from fnd_user where user_name='ORACLE';

  select organization_id

    into l_organization_id

    from hr_all_organization_units

    where name='Vision Operations';

  fnd_global.apps_initialize(l_user_id,l_responsibility_id,l_application_id);  

  --mo_global.set_policy_context('S',l_organization_id);

  mo_global.set_policy_context('M',l_organization_id);

--  dbms_output.put_line('Done...');

end;

Concurrent Requests

Concurrent request can now be setup to run as in 11i (default) or as MOAC compliant programs.

Use the “System Administrator” OAF screen “Concurrent Program” to set Operating Unit Mode to Single (OU) or Multi (MOAC compliant):

You can either run out and buy a huge server and re-install on that or try to use two laptops instead of one?

Equipment used:

• Router – Draytek 2820
• Laptop 1 – R12 server: Lenovo Edge 13 with AMD Turion Neo X2 CPU and 4Gb RAM
• Laptop 2 – Client: Dell XPS M1530 with Intel T8300 CPU and 6Gb RAM
• Hard disk – R12 disk: Akasa eSata/USB2.0 case with Seagate 500Gb 7200RPM

Follow this to make it work:

1. In the router set a static IP address for laptop 1 – I use 192.168.1.4
2. Connect the hard disk to laptop 1 and boot
3. So laptop 1 still thinks it is 127.0.0.1 but router makes it 192.168.1.4 for other computers
4. Start R12 on laptop 1 and ensure everything is up and running nicely
5. Update hosts file on laptop 2:
   192.168.1.4 localhost.localdomain
   Localhost is normally the loopback but as the R12 server is using this name we must do the same
6. In the browser on laptop 2 connect to http://localhost.localdomain:8000
7. Login and go!

Conclusion

R12 Performance is fine only taking 1.7Gb RAM on Laptop 1.

Cpu level is very low as well < 25% overall.

I/O could be better as the USB2 is not that fast.

Fitting the R12 hard disk inside Laptop 1 may boot performance thanks to Sata being 10+ times faster than USB2.

However it in large scale projects you may have to submit very large request sets depending on some logic which is virtually impossible to implement as a Concurrent Request Set.

It is easy to submit a program from a PL/SQL program using:

FND_REQUEST.SUBMIT_REQUEST

However how do we keep track of all the submitted sub-programs in similar way as a normal concurrent request set do it – retaining the parent request id?

So what we want to have is:

> Request Set
>> Request Job (parent request id = Request Set)

Also how do we ensure that we do not submit too many sub-programs at any one time in order not to clutter the request queue?

So what we really want to have is:

> Request Set
>> Request Job (parent request id = Request Set)
> Wait for child requests to complete
>> Request Job (parent request id = Request Set)

The problem is that when a parent program wakes up it re-submits itself. So in order to make this work you need to use recursive programming methods.

The trick is to use these three calls:

• FND_CONC_GLOBAL.SET_REQ_GLOBALS with Status=PAUSED and REQUEST_DATA
  Status = PAUSED will put the parent program into a wait state until child processes have completed
  REQUEST_DATA parameter will save data between runs
• FND_CONC_GLOBAL.REQUEST_DATA
  Retrieves data from previous run
• FND_REQUEST.SUBMIT_REQUEST with SUB_REQUEST=>true
  Submits the sub-programs with control of a parent program.
  If set to “false” you have no control of the sub-programs.
  If you set this to “true” but without setting parent program status to PAUSED you will get the issue where child programs are phase=“inactive” and status=“no manager”

So in my test I will create to sets of two jobs:

> Request Set
> Set 1:
>> 1: Request Job (parent request id = Request Set)
>> 2: Request Job (parent request id = Request Set)
> Wait for child requests to complete
> Set 2:
>> 1: Request Job (parent request id = Request Set)
>> 2: Request Job (parent request id = Request Set)

So running my concurrent request set:

Set 1:

So two first programs has just run and parent quest is about to start the second job set.

Set 2:

Note all jobs has the same Parent Request ID.

Nice and controlled.

Create the PL/SQL code:

Package:

create or replace
package XXKWICONCSET authid current_user as
  — this is exposed for conc manager
  procedure XXKWI_REQUEST_SET_CONCUR (
    ERRBUF OUT NOCOPY varchar2,
    RETCODE OUT NOCOPY varchar2
  );
  procedure XXKWI_REQUEST_CONCUR (
    ERRBUF OUT NOCOPY varchar2,
    RETCODE OUT NOCOPY varchar2,
    PARAM IN varchar2
  );
end XXKWICONCSET;
/

Package Body:

create or replace
package body XXKWICONCSET as
  — Concurrent child program
  procedure XXKWI_REQUEST_CONCUR (
    ERRBUF OUT NOCOPY varchar2,
    RETCODE OUT NOCOPY varchar2,
    PARAM IN varchar2
  ) is
  begin
    fnd_file.put_line(fnd_file.output,’Running job: ‘||param);
    ERRBUF:=’Done Job…’;
    RETCODE:=0;
  end XXKWI_REQUEST_CONCUR;
  — Concurrent Request Set
  procedure XXKWI_REQUEST_SET_CONCUR (
    ERRBUF OUT NOCOPY varchar2,
    RETCODE OUT NOCOPY varchar2
  ) is
    L_REQUEST_ID number;
    L_REQ_DATA varchar2(10);
    job_count number;
    set_count number;
  begin
    L_REQ_DATA:=FND_CONC_GLOBAL.REQUEST_DATA; — get previous job data
    if L_REQ_DATA is null then — first run set count to ‘0′
      L_REQ_DATA:=’0′;
    end if;
    set_count:=to_number(l_req_data)+1; — increment for every job
    fnd_file.put_line(fnd_file.output,’Running set: ‘||to_char(set_count));
    if set_count>2 then — stop after 2 jobs
      ERRBUF:=’Done Set…’;
      RETCODE:=0;
      return;
    end if;
    — submit 3 child programs
    for job_count in 1..2 loop
      fnd_file.put_line(fnd_file.output,’Submitting job: ‘||to_char(job_count));
      L_REQUEST_ID:=FND_REQUEST.SUBMIT_REQUEST(
        APPLICATION=>’FND’,
        program=>’XXKWI_REQUEST_CONCUR’,
        DESCRIPTION=>to_char(job_count),
        START_TIME=>null,
        SUB_REQUEST=>true, — True = This is a child program
        ARGUMENT1=>to_char(job_count) — data to job
      );
    end loop;
    — Wait until done and save set count:
    FND_CONC_GLOBAL.SET_REQ_GLOBALS(
      CONC_STATUS=>’PAUSED’, — wait for child request to complete
      REQUEST_DATA=>to_char(set_count) — save set data
    ); 
    Commit;
  end XXKWI_REQUEST_SET_CONCUR;
end XXKWICONCSET;
/

Register the two executables and programs.

Concurrent Job:

Concurrent Set:

Request Group:

So I downloaded R12.1.1 from Oracle eDelivery. 40 files in excess of 45Gb to the joy of my ISP.

NOTE until 20/11-2009 there was a corrupt file on eDelivery and it took me a while to convince Oracle that it was the case.

The sick file was: V15571-01_2of3.zip

However the file has been fixed after the above date but be sure to test your downloaded files anyway using ziptest.

Otherwise follow the procedure for R12 Installation.

Some of the initial rapidwiz screen has changed though.

The Welcome screen is saying 12.1.1 – great:

Select express install:

Just click next on this one:

Update base directory:

Before logging in you should compile the jsp pages – see how to Compile JSP.

And then it is business as usual like in R12.0.4

Also in you environment script – the one I call vis.sh change this:

from:
TECHENV=/oracle/VIS/db/tech_st/10.2.0/$CONTEXT_NAME.env
to:
TECHENV=/oracle/VIS/db/tech_st/11.1.0/$CONTEXT_NAME.env

Have fun…

This approach is recommended when downloading huge amounts of data like when downloading Oracle R12 installations files.

In addition to testing the downloaded files also be sure to use a download manager.

The basic zip test command is:

$ unzip –t –qq {zipfile}

I made a Linux script for it named ziptest.sh:

A similar script can be made for Windows names ziptest.cmd:

As command line programming in windows isn’t great we need to call a subprogram names ziptest_call:

That should be it.

Be sure to update directory references for your own purpose.

Then add your Apps server to IE8 security settings:

Lower the security setting to medium-low for Trusted Sites:

Allow pop-ups from the Apps server:

When you start the application for the first time you may get one or more warnings like this:

Just keep clicking run and it should work.

Note IE8 is not yet supported by Oracle.

I’m running R12.0.4 on OEL5.3 with Virtualbox 2.2.2 having a Windows 7 machine.

R12

R12 is installed on OEL with the hostname: localhost.localdomain

Firefox works locally against R12 so the EBS stack in working at this point.

Start the database, listener and apps. See Managing R12:

$ addbctl start

$ addlnctl start VIS

$ adstrtal apps/apps

To test if a database network connection would work try:

$ sqlplus apps/apps@VIS

Virtualbox

Installed Vbox 2.2.2 on OEL. Get it from http://www.virtualbox.org

Setup Vbox host-only network in \File\Preferences

and give it a ip like 192.168.2.7 on the adapter tab:

and setup the dhcp tab like this:

When the Vbox machine is started you should be able to ping it:

Windows 7 in Virtualbox

I use Windows 7 RC.

Create a Vbox machine and mount the downloaded Win7 ISO as CDROM and run install:

You get it from here:

Setup Vbox networking like this:

This will communicate through the 192.168.2.7 ip address.

When Window 7 is installed create a new entry in the file:

C:\windows\system32\drivers\etc\hosts

like this:

Pointing to localhost.localdomain so Firefox R12 login works.

To ensure everything work try to ping localhost:

SQLDeveloper

If you want to do development then this tool is great. Get it from:

http://www.oracle.com/technology/products/database/sql_developer/index.html

I installed version 1.5.1.

Setup the connection:

I just used the ip address here and it works fine.

Firefox

To run R12 forms and framework you need Firefox 2.

Install Firefox 2 in Win7. I installed version 2.0.0.14. Get it from:

http://www.mozilla.com/en-US/firefox/all-older.html

Allow localhost.localdomain is Firefox security:

and pop-ups:

To start R12 enter the URL:

http://localhost.localdomain:8000/OA_HTML

And you get the login screen:

Java

Java is a must for R12. Check the version required when you have logged in to R12 and click a forms screen

In 12.0.4 it is 1.5.0.13:

Then install Java 1.5.0.13 to ensure you don’t get plug-in issues with R12. Get it from:

http://java.sun.com/products/archive/j2se/5.0_13/index.html

As Firefox only supports one Java plug-in at any one time check the directory for plug-ins:

C:\ > DIR “C:\Program Files\Mozilla Firefox\plugins”

Ensure you only have one Java plug-in here. If it is missing the do this:

C:\ > copy C:\Program Files\Java\jre1.5.0_13\bin\NPJPI150_13.dll C:\Program Files\Mozilla Firefox\plugins

Restart Firefox and now everything should work.

OEL does not seem to be a complete operating system – it feels like Linux 5 years ago with loads of drivers missing and command line fiddling.

One chapter for each part…

Graphics

This does work out of the box but the built in driver is not great and make the fan run at full all the time.

If you installed in Virtualbox then fix the graphics card by getting this driver:

http://www.nvidia.com/object/linux_display_ia32_180.51.html

You have to install it outside X11 so issue this command:

$ init 3

To exit X11 and login to a command prompt.

Then run the Nvidia install script:

$ ./NVDIA*run

When done then:

$ init 5

To get X11 back.

Network

Should work. Ensure that \System\Administration\Network has a entry using Marvell network driver.

Wireless

I configured the network is two steps.

First ensure you got a wlan0 device in \System\Administration\Network in Devices and Hardware tab.

Next ensure you have define wlan security in \System\Preferences\More Preferences\Network Connections in Wireless tab.

After all this wireless still doesn’t work but try these two lines:

$ chkconfig NetworkManager on
$ service NetworkManager start

This starts the wireless service and now it should work.

Bluetooth

Bluetooth does not work immediately – for some reason it does not detect any devices. You normally get this:

$ hcitool scan
Scanning …
Inquiry failed: Connection timed out

First reset Bluetooth to enable discovery:

$ hciconfig hci0 reset

If the Bluetooth does not discover after reboot try to add the reset to /etc/modprobe.con

options hci_usb reset=1

Then scan for any devices:

$ hcitool scan
00:1D:D8:98:6F:23       Microsoft Bluetooth Notebook Mouse 5000

Bingo – things start to work now…

Next configure the security settings by editing /etc/bluetooth/hcid.conf and change to (in bold):

Then ensure the Bluetooth manager discovers the device by editing /etc/default/bluetooth:

Then restart Bluetooth:

$ /etc/init.d/bluetooth restart

Connect to the new device using:

$ hidd –search
Searching …
Connecting to device 00:1D:D8:98:6F:23

And everything should work now!!!

At least my MS BT mouse did except for the scroll wheel.

MS BT Mouse – Microsoft Bluetooth Notebook Mouse 5000

First get the Bluetooth to work as above.

Touchpad

When touching my touchpad it is erratic.

First get hold of your input devices.

Probe the mouse devices using this command:

$ udevinfo -a -p `udevinfo -q path -n /dev/input/mouse1`

And change the mouse# until you get an output like:

The ALPS is your touchpad.

Update your /etc/X11/xorg.conf to include mouse1:

Now we need to fix the erratic behaviour.

Edit /boot/grub/menu.lst and include “i8042.nomux=1” when starting Linux:

Reboot and the touchpad should work now.