Oracle Applications R11.5.2 Installation/Upgrade Issues

Fields of interest

Note:

This page contains the 11.5.2 issues originally documented under Note 112659.1. Since 11.5.3 is now generally available, the orginal note will only contain 11.5.3 issues. This page will no longer be maintained and may therefor contain obsoleted information.

Dutch Support installation policy

As a result, the policy of Dutch Support is to install all the components including Webserver and Formsserver on the Database Server, even though it will not be used. This way we will prevent problems in both adpatch and adaimgr.

General R11i install issues

On some platforms we found, that the 11.5.2 fresh install database did not contain the java classes for the xmlparser. This caused us to run into bug 1336941. The solution is to load the xmlparser and the Xsu12 as described in the 'finishing your installation' section of the 11.5.1 installation manual.

If you are installing on a system that has an existing Oracle installation which has been installed using the Oracle Universal Installer, then you might have some permission problems. The Oracle Universal Installer creates a file called oraInst.loc in the /var/opt/oracle or /etc directory which contains the PATH to your 'software inventory'. The account you are using to install must have WRITE permissions on this directory and on ALL directories and files within this directory.
If it doesn't then the Rapidwiz will be unable to start the Oracle Universal Installer (without giving any error).

The manual incorrectly states that the FRESH installed database is 8Gb. In reality it's 11Gb !!
This issue has been raised and the documentation will be fixed for the 11.5.3 release.

In 11.5.2, the start and stop scripts for the database and 8.1.6 listener are no longer located under the <COMMON_TOP>/admin/scripts. They are now stored under the <ORACLE_DB_DIR>/8.1.6/appsutil/scripts directory.

If you are distributing the software and database over multiple disks, then the Universal Installer will probably tell you you don't have enough diskspace. This is not true. Be sure to check our sizing page to find out the actual sizes for the different components.
Click here to find the dutch APPS sizing page (NOT AVAILABLE VIA METALINK!!).

Make sure to always enter a value for the DNS name in the rapidwiz.
Failing to do so will give problems especially in the CRM area, where a domain is expected.

If for some reason the Oracle Universal installer seems to be ignoring the TMPDIR setting. As a result all temp files will be written to /tmp. Make sure you have about 650Mb of space free in /tmp.
Alternative solution  is to set the variable TMP which will be picked up by the Oracle Universal Installer. See note 2062913.102 for this.

Some of the CD packs which were shipped from Ierland do not have the right title printed on the CD. This does not apply to CD packs shipped from the US..
The CD's are called CD 1 of XX. The rapidwiz will no longer ask for CD 4, it will ask for the 'ora 8i' CD.
This should be fixed by the end of november 2000.

If you have a CD pack like this, then you either have to check the CD labels by mounting them, or just try the next CD.

On some platforms the universal installer will ask you to enter the JDK Home (default answer is /opt/java).
If this question is being asked, then the rapidwiz will not deliver the required JDK1.1.8 software and you have to install it on the system.
On Solaris and NT, the JDK is being installed by the rapidwiz.

The directory you specify here should contain a bin directory with the java executable in it.
If you want to download te JDK, then you can do so using the following URL's:

• IBM AIX-Based Systems: http://www.ibm.com/java/jdk/aix/index.html
• Compaq Tru64 UNIX: http://www.digital.com/java/download/index.html
• HP 9000 Series HP-UX: http://www.unix.hp.com/java/java1/jdk_jre/downloads/index.html
• Linux Intel: http://www.ibm.com/java/jdk/118/linux

General R11i which issues occur after completing the installation/upgrade

Some platforms (like HP) do not have the unzip utility located in the <common_dir>/util.
This can be a problem if you want to apply patches.

In order to be able to relink files, you must apply patch 1313962 or better, apply the cummulative AD patch (this is currently 1553656).
Before doing so, make sure that adrelink.sh and adjrebuild.sh have the x-bit set. If not run the following command:

    chmod +x $AD_TOP/bin/adjrebuild.sh
    chmod +x $AD_TOP/bin/adrelink.sh

Based on bug 1404689 it's very well possible that the database installed by rapidwiz has statistics setup for the tables under the SYS account.
This is not supposed to be the case as this will cause performance issues rather then solve them.
To check whether the SYS schema has been analyzed, run the following 2 SQL statements:

select table_name
from dba_tables
where owner = 'SYS' and  last_analyzed is not null
/
select index_name
from dba_indexes
where owner = 'SYS' and  last_analyzed is not null
/

execute dbms_stats.delete_schema_stats('SYS')

When you have completed you install  upgrade you will have to create a digital certificate using 'adjkey -intialize' as described in 'Finishing Your Installation' section of the installation manual
To prevent the yellow waring bar (Warning: Applet Window), please follow the following precedure:

• Run adjkey -initialize to create the digital certificate
• Run adjbuild.sh to repackage the jnitiator
• Run adadmin and generate ALL jar files (unconditionally = yes)
• Install jinitiator on a PC.
• Reboot the PC (just to be sure)

If you plan to share the digital certificate over multiple 11i instances, then please check note 119322.1 about which files to copy.

R11i upgrade issues

Export/Import vs. Migrate utility

When using the migrate utility, you will need to do the following preparations:

• System tablespace should have more than 600Mb free.
• Modify the system rollback segment with the following statement: ALTER ROLLBACK SEGMENT SYSTEM STORAGE (NEXT 1M);

When using import/export to upgrade to 8.1.6, make sure that the export is run with the parameter ‘COMPRESS=Y’.
Before running the import, the 8.1.6 database has to be created with all tablespaces, datafiles and rollback segments.
An upgrade on NT revealed, that the IMP utility will not create datafiles with a maximum size of  1GB which could cause the import to fail if you have datafiles that are larger then 1Gb..
Please note, that the Rapidwiz will not deliver database create scripts. So if you want to do an export/import, you manually have to create the scripts to build the database.

Suggested actions before running the R11i upgrade (after unloading the R11i software):

Based on the 10.7 to 11i upgrade you will need the following space available in the 10.7 database before the upgrade (for release 11 this will be less):

Note that these are the minimum figures:

SYSTEM	1600Mb + 700Mb if MRC will be used
ROLLBACK	200Mb
TEMP	200Mb
USER_DATA	800Mb
USER_INDEX	900Mb

Check out the Apps related RDBMS issues for any RDBMS issues you might encounter.

NLS customers that wish to upgrade from 10.7 NLS to 11i US have to deactive and delete the language before upgrading. In order to do so you have to run the following 2 scripts:

• nlsdeact.sql
• nlsmldel.sql

If you do not delete the language before the upgrade to 11i, then you have to upgrade from 10.7 NLS or 11.0.X NLS to 11i NLS as described in the 11i NLS release notes. Once the upgrade is completed you can revert back to US as described in these release notes. Obviously it's faster to delete the language up front.

Before upgrading (after unloading the software) you should apply the latest cummulative AD patch (this is currently 1553656).
This patch should be applied in preinstall mode preferable before doing the R11i upgrade. Make sure to read the readme.txt.
This patch will also fix a problem if you have any customized schema names (e.g. APPS_FND for the APPS account).

Be aware of the fact that while applying this patch you might find that the X bit is missing from the adrelink.sh and adjbuild.sh script in the $AD_TOP/bin directory. Workaround is to do the following:

• chmod 755 $AD_TOP/bin/adrelink.sh
• chmod 755 $AD_TOP/bin/adjbuild.sh

Before upgrading you should apply the high priority ARU's related to upgrades.

NOTE !!:

Due to the rapid releasing and obsoleting of patches, the list below might be out of date.
We are doing out best to keep up and update this list frequently, but you might find some patches that have been obsoleted. If so, please make a remark against this note so we can change this.

(*) Click here for additional information about NLS preupgrade patches

The patches needed by customers doing an Order Management upgrade are now all included in the Consolidated MFG preupgrade patch.
There is no need anymore to install patches 11.5.0 upgrade, but before running the dbupg11i.drv.

Rapidwiz will incorrectly setup intermedia. As a result scripts like icxintm2.sql will error out with errormessages like:
   sqlplus -s APPS/APPS @/u01/applsean/11i/icx/11.5.41/admin/sql/icxintm2.sql
   ICX ICX Connected.
   DECLARE
   *
   ERROR at line 1:
   ORA-29855: error occurred in the execution of ODCIINDEXCREATE routine
   ORA-20000: interMedia Text error:
   DRG-50704: Net8 listener is not running or cannot start external procedures
   ORA-28575: unable to open RPC connection to external procedure agent
   ORA-06512: at "CTXSYS.DRUE", line 126
   ORA-06512: at "CTXSYS.TEXTINDEXMETHODS", line 54
   ORA-06512: at line 1
   ORA-06512: at "SYS.DBMS_SYS_SQL", line 782
   ORA-06512: at "SYS.DBMS_SQL", line 32
   ORA-06512: at line 30

For details see bug 1314677 and NOTE 97824.1.

You have to setup intermedia my modifying the 8.1.6 listener.ora. It should contain the following information (changes are highlighted):
   LISTENER =
    (DESCRIPTION_LIST =
     (DESCRIPTION =
       (ADDRESS_LIST =
        (ADDRESS = (PROTOCOL = IPC)(KEY = EXTPROC0) ) )
          (ADDRESS_LIST =
          (ADDRESS = (PROTOCOL = TCP)(HOST = ap118sun)(PORT = 1521))
          ) ) )

    Note: EXTPROC0 ends in a zero.

Also a system identifier (SID) name of PLSExtProc and a program name of EXTPROC in the server's LISTENER.ORA file.

 SID_LIST_LISTENER =
  (SID_LIST =
    (SID_DESC
      .....
    (SID_DESC=(SID_NAME=PLSExtProc)
    (ENVS=LD_LIBRARY_PATH=/usr/dt/lib:/oracle/prodora/8.1.6/rdbms/lib:/oracle/prodora/8.1.6/lib:/oracle/prodora/8.1.6/ctx/lib)
    (ORACLE_HOME=/oracle/prodora/8.1.6)
    (PROGRAM=extproc))

Note !!! The ENVS section no longer needs the quote's '' around the variables like in RDBMS 8.0.

Add an entry for EXTPROC0 in the tnsnames.ora file you are using with adaimgr. This is very likely the 8.0.6 tnsnames.ora, but you might want to check the TNS_ADMIN variable for this:
For example, add this to the end of tnsnames.ora:

  extproc_connection_data =
   (DESCRIPTION=
    (ADDRESS_LIST =
     (ADDRESS=(PROTOCOL=IPC)(KEY=EXTPROC0))
     )
     (CONNECT_DATA=
       (SID=PLSExtProc)
        (PRESENTATION = RO)
      ) )

Note !!! EXTPROC0 ends in a zero, RO ends in a letter O (not a zero).

When finished restart the 8.1.6 listener and test the extproc connection using 'tnsping'.

Note !!! The Rapidwiz might have defaulted the extproc entry in the listener, but has probably named in extproc<$ORACLE_SID.

During the R11i upgrade we found that for some reason indexes existed under the APPS schema which were supposed to exist under the same schema. As a result adodfcmp will fail to create the index under the base account as there is already an existing index on the same columns.
The solution here is to drop the index under the APPS schema and restart the script.

Example indexes where we found the problem:
    MTL_CUSTOMER_ITEMS_U2
    CHV_SCHEDULE_HEADERS_N7

If as, as part of the upgrade preperation, the 'Consolidated 11i upgrade fixes for the Oracle Financials Family' patch 1660557 or 1660907 has been applied,  then there are some issues that need to be fixed (specifically for AP performance patch 1411882 which is included in this consolidated upgrade fix).
This patch requires the backport of RDBMS patch 1253654 to be installed. This is to fix a problem with incorrect result. Failing to apply this patch could cause data corruption. The backport numbers are:
 

OSF1	1366904
WINDOWS_NT	1253654  (requires 8.1.6.1.2 to be installed first)
IBM_RS6000	1351677
HP-UX	1294287
SOLARIS	1289110

The readme.txt of this patch mentioned changes that are needed in the init.ora. We have interpreted these to be:

• HASH_AREA_SIZE = <between 512000 and 30Mb>
• PARALLEL_MAX_SERVERS = 60

The size of HASH_AREA_SIZE depends on the system memory. Here are the settings we believe are correct:

• System with 1Gb memory, HASH_AREA_SIZE should be set to 5Mb or less
• System with 2Gb memory, HASH_AREA_SIZE should be set to 20Mb or less
• System with 3Gb or more, HASH_AREA_SIZE can be set to 30Mb

Make sure that the redo log files are at least 20Mb or larger.
Make sure that you have sufficient TEMP tablespace (this can be up to 3Gb or more), or turn autoextend on using the following command:

    alter database tempfile '<temporary datafile>' autoextend on next 100M;

If PA is installed and there are multiple Operating Units defined using the same Set of Books, then you could very well run into bug 1345756.
As a result the pa111u05.sql and pa111u06.sql will fail with the following error:

pa111u05
declare
*
ERROR at line 1:
ORA-20001: Updating EI table,min_id= 156465 max_id=312868:ORA-06512: at line1
ORA-06512: at line 172

paa11u06
declare
*
ERROR at line 1:
ORA-20001: Updating CDL table, min_id= 61 max_id=625674:ORA-06512: at line 1
ORA-06512: at line 158

Since there is no fix for this problem yet, the workaround is to modify following part of the pa111u0[56].sql scripts:
         IF ( FIRST_RECORD ) THEN
           l_prev_sob_id := org_rec.sob_id;
            dbms_application_info.set_client_info(org_rec.org_id);
            pa_parallel.G_Curr_Tab(nvl(org_rec.org_id,-99)) :=
                 pa_currency.get_currency_code;
            FIRST_RECORD := FALSE;
         END IF;

Comment out the line "FIRST_RECORD := FALSE;" so that the code will get executed for all the OU's not just for the first record.

Important note for customers upgrading Order Entry !!

Customer running Order Entry are advised to check the document under Note 121200.1. This document describes how to upgrade Order Entry to Order Management in 11i.

Version 3.0 of this document dated November 2000 does not contain the correct patch information.
Here is additional information for upgrading 11.5.2 customers:

• Patch 1323676 (page 8) is included in 11.5.2 and do not have to be applied when upgrading using 11.5.2 software.
• Patch 1471586 (page 9) has been replaced by patch 1561685.
• Patches 1311832,1324853, 1324883 (page 9) are included in 11.5.2 and do not have to be applied when upgrading using 11.5.2 software.
• Patch 1310156 (page 9) is included in 11.5.2 and does not have to be applied when upgrading using 11.5.2 software.

• Patch 1612819 listed on page 4  has been replaced by patch 1667156.
• Patch 1612824 also listed on page 4 has been replaced by patch 1667152.
• Patch 1323676 listed on page 6 is included in 11.5.2 and 11.5.3 and does not have to be applied when upgrading using 11.5.2 or 11.5.3 software.

Customers upgrading from 10.7 to 11.5.2 can receive the following error while running adaimgr:
AutoUpgrade error:
The following ORACLE error:
.
ORA-1400: cannot insert NULL into
("APPLSYS"."FND_APPLICATION"."APPLICATION_NAME")
.
.
occurred while executing the SQL statement:
.
INSERT INTO FND_APPLICATION
 (APPLICATION_ID,APPLICATION_SHORT_NAME,LAST_UPDATE_DATE,LAST_UPDATED_BY,CR
@ EATION_DATE,CREATED_BY,LAST_UPDATE_LOGIN,BASEPATH) VALUES
 (602,'XLA',to_date('01/01/1980','DD/MM/YYYY'),1,to_date('01/01/1980','DD/M
M/YYYY'),1,0,'XLA_TOP')

If you fix this, then the next error you will get is the fact that the SOURCE_LANG column isn't present in FND_APPLICATION_TL table.
This issue has been raised in bug 1536306, and is fixed in the latest cummulative AD patch (this is currently 1553656).

Issues found in the pre-upgrade steps for 11i:

Step 1 of the Database Upgrade steps (Category 2 - After you receive the software) will let you run the advrf107.sql script. This script showed some packages having lines > 255 characters. Packages having lines > 255 characters will cause a problem when running Maintain MRC schema's in adadmin.
The script therefore instructs you to change these packages. Standard APPS packages mentioned here can be ignored as they will all be replaced by the upgrade.

Step 1 of the Oracle Flexbuilder/Account generation tasks (Category 2 - After you receive the software) will let you run the afffcfb.sql script.
This script produced a large number of problem records where the SEGMENT USAGE was NULL. The instructions in the output file tell you to change this in the flexbuilder assign screen. The problem we found was that a large number of records were actually protected flexbuilder assignments. Due to these protection we could not change them. We therefore assumed this was seeddata which did not need to be changed.

Also note, that if the output tells you to run the FNDNLINS.sql script, then the script you need if the R10.7 or R11.0.3 version of the script.
The 11.5 version of the script requires 11.5 packages which you don't have at this point in time.

Step 4 of the Database and System Administration Upgrade Steps (Category 3 - Performing the upgrade) tells you to run addb816.sql.
When running the addb816.sql script you could encounter the following problem duing the 'create or replace java system;' statement:
    ORA-4030 out of process memory

Current workaround for this problem is the following:

• Run the rmjvm.sql script to remove possible left-overs.
• Start up svrmgl and login as internal
• Run the statement: "create or replace java system;"  (This should now succeed without errors.)
• Comment out the above create statement in the initjvm.sql script and
• Re-run the addb816.sql script.

Step 4 of the Database and System Administration Upgrade Steps (Category 3 - Performing the upgrade) tells you to run addbctx.sql.
The manual mentions that on Unix you should run the script with the following parameters:

Example of running addbctx.sql (on Unix):
    sqlplus system/manager @addbctx.sql FALSE CTXSYS TEMP $ORACLE_HOME/ctx/lib/libctxx8.so

There are two pitfalls in this:

• It is not allowed to use environment settings like $ORACLE_HOME, you MUST specify the complete path to the 8.1.6 ORACLE_HOME!
• Don't just specify the library file as 'libctxx8.so' but check this. Under Solaris the file is called libctxx8.so, under HP/UX 11.0 it is called libctxx8.sl.

Step 5 of the Database and System Administration Upgrade Steps (Category 3 - Performing the upgrade) tells you to run adwdb25.sql.
On HP we found that this script contained hardcoded path settings which we incorrect. It's possible this problem exists on other platforms as well.

Step 6 of the Database and System Administration Upgrade Steps (Category 3 - Performing the upgrade) tells you to run admodpls.sql.
On some platforms, the files which are nun by this script are not located under the Apache/modplsql/owa directory, but onder the webdb30/modplsql/owa directory. This is documented under bug 1343287.

Before starting adaimgr, make sure that your ORACLE_HOME is set to 8.0.6, not 8i !!

When adaimgr is verifying the files needed for upgrade, you might run into an issue where some PER files in the html directory are missing. For example:

• html/context_layer_pos.js
• html/dhtmlapi.js
• html/hrchsupw.js

    $ cd $GHR_TOP/patch/115/sql
    $ sqlplus apps/apps @ghgenral.pkh
    $ sqlplus apps/apps @ghgenral.pkb

This will create the GHR_GENERAL function which is needed by packages which is used in the mentioned triggers.

While running an upgrade from 10.7 to 11i, you could encounter the following error while running afpnls01.sql:

ORA-00001: unique constraint (APPLSYS.FND_APPLICATION_TL_U2) violated occurred while executing the SQL statement:
UPDATE FND_APPLICATION_TL T setLANGUAGE = ( select L.LANGUAGE_CODE from
 FND_LANGUAGES L where L.NLS_LANGUAGE = T.LANGUAGE ) where exists ( select
 L.LANGUAGE_CODE from FND_LANGUAGES Lwhere L.NLS_LANGUAGE = T.LANGUAGE )

Based on the information from previous upgrades, the problem records in FND_APPLICATION_TL always seem to be:

APPLICATION_ID LANGUAGE   APPLICATION_NAME
-------------- --------------------------------------------------
710            AMERICAN   Oracle Release Management
662            US         Oracle Release Management
711            AMERICAN   Oracle Automotive
663            US         Oracle Automotive

The afpnls01.sql wil try to update the LANGUAGE from AMERICAN into US, and thus causing  the ORA-1 on FND_APPLICATION_TL_U2 as this is an index on the columns (APPLICATION_NAME, LANGUAGE ).

Solution, based on fresh installed 11i database is the following:

update fnd_application_tl
set application_name = 'Oracle Release Management Integration Kit'
where language = 'AMERICAN' and application_id = 710
/
update fnd_application_tl
set application_name = 'Oracle Release Automotive Integration Kit'
where language = 'AMERICAN' and application_id = 711
/

After these updated the afpnls01.sql should run through without problems.

Problems when running the dbupg11i.drv driver:

During an 10.7 to 11.5.2 upgrade we found that the dnupg11i.drv failed running WFLOAD for iemwfprc.wft .
The error we received in the logfile was:

Report filename : /appl/df11/df11appl/admin/DF11/out/ANONYMOU.806128
wferr:
  - 1602: Could not save.
  - 1400: Could not save to database. MODE=UPLOAD EFFDATE=<null>
  - 1404: Please first load this entity or check protect and custom level of LOOKUP_TYPE 'WFSTD/CRASH_STATUS'.
Oracle Workflow Definition Loader 2.5.0
Access level: 20, Mode: UPLOAD
  ITEM_TYPE 'WFSTD' is protected, no changes were saved.
  ITEM_ATTRIBUTE 'WFSTD/OWNER_ROLE' is protected, no changes were saved.
  ITEM_TYPE 'WFMAIL' is protected, no changes were saved.
  LOOKUP_TYPE 'WFSTD/CRASH_STATUS' is protected, no changes were saved.

The workaround is  update the $IEM_TOP/11.5.0/patch/115/import/US/iemwfprc.wft and set the ACCESS_LEVEL=0.
You could still get a problem with

• $GMD_TOP/11.5.0/patch/115/import/US/gmdnprdd.wft
• $GML_TOP/11.5.0/patch/115/import/US/GMLPUPAY.wft

To prevent this problem, you can install patches 1506963 and 1519368.

Patch 1519368 will cause WFLOAD to continue if it cannot load specific item types. These item types should reside in fndwfstd.wft. However if you have fndwfstd.wft 115.42, then the item types being skipped do not exist at all. Therefor you have to install patch 1506963 as well.

During an upgrade where patch 1542052 (latest version of dbupg11i.drv) had been applied, we found that problems similar to bug 1498734 and note 125959.1 occured wile running csdscursp.ldt.
The error we received was:

Altering database NLS_LANGUAGE environment to AMERICAN
Uploading FND_APPLICATION CSD
Uploading FND_RESPONSIBILITY CSD DEPOT REPAIR
A database error occurred:
ORA-20001: Invalid foreign key data was found for this record.
Value CSD_MAIN_MENU for column MENU_NAME does not exist in table
FND_MENUS_VL.
ORA-6512: at "APPS.APP_EXCEPTION", line 47
ORA-6512: at "APPS.FND_RESPONSIBILITY_PKG", line 215
ORA-6512: at line 10

$FND_TOP/bin/FNDSLOAD APPS/FND 0 Y CSD patch/115/import/US/csdmenu.slt UPLOAD

The dbupg11i.drv could fail running the following command:

FNDLOAD APPS/APPS 0 Y UPLOAD @INV:patch/115/import/invicset.lct @INV:patch/115/import/US/invascst.ldt -

Uploading from the data file /appl/df11/df11appl/inv/11.5.0/patch/115/import/US/invascst.ldt
Altering database NLS_LANGUAGE environment to AMERICAN
Uploading MTL_CATEGORY_SETS 5
A database error occurred:
  ORA-06550: line 21, column 1:
PLS-00306: wrong number or types of arguments in call to 'LOAD_ROW'
ORA-06550: line 21, column 1:
PL/SQL: Statement ignored

During the upgrade of an Order Management customer we encountered an ORA-0001 in ontup206.sql:

ora-1 ontup206.sql unique constraint
ora-6512 ont.oe_price_adjustments_u1

Known issues during the upgrade of a global demo database

See the following bugs:
    1297177
    1209613

Be aware that the global demo database has some corrupted data, so the upgrade will fail several times.

Issues found in the post-upgrade steps for 11i:

Step 2 of the Database Upgrade tasks (Category 4 - Before using Oracle Applications) will let you run Rapidwiz to reconfigure the APPS install. The rapidwiz will unload the remainder of the configuration files and will configure them.
In this specific upgrade, the FNDNAM was set to APPS_APPDEMO, not APPS. After running the configure step however we found that some files were configured as if FNDNAM was set to APPS.
It is possible that this was caused by some environment setups, but it's worth while checking whether these files are configured correctly.

The following files were configured incorrectly:

• <common dir/portal/logon_s.html
• <common dir/portal/apccmsvs.htm
• <common dir/html/bin/appsweb.cfg
• <common dir/admin/install/adgendbc.sh
• <common dir/admin/install/adsvcm.sh

Step 4 of the Database Upgrade tasks (Category 4 - Before using Oracle Applications) will let you run the adupdprf.sql script to update the profile options.
Please check out Note 125442.1 and bug 1505552 which explain this script is not correct.

If you run this script, then signing on to the Personal Homepage will ive the following error message:
    The requested URL /PROD/OracleConfigure.render was not found on this server.

The solution is to run the adupdprf.sql from the 8.1.6 ORACLE_HOME under the appsutil/install directory.

Step 1 of the Applications Implementation wizard (Category 5 - Before using your products) tells you to run 2 scripts. The example in the manual however does not show any parameters. This is not true. If you read the paragraph below the examples you will see a parameter is required though.
The correct syntax is:
    $ cd $AZ_TOP/admin/sql
    $ sqlplus <APPS username/<APPS password @azwardet.sql %
    $ sqlplus <APPS username/<APPS password @azwarsum.sql % Y

Step 2 of the Oracle Payables tasks (Category 6 - Before using product features) tells you to run apxfdp01.sql. This script has following syntax:

    sqlplus <APPS username/<APPS password @apxfdp01.sql <batchsize

The batchsize parameter is not documented.

Step 1 of the System Maintenance Upgrade Task (Finishing your upgrade) tells you to run podold.sql. This script  has the following syntax:

    sqlplus <APPS username/<APPS password @$PO_TOP/upgrade/sql/podold.sql \
    <PO username <PO passwordAPPS username <APPS password

The correct syntax is:
    sqlplus <APPS username/<APPS password @$PO_TOP/admin/sql/podold.sql \
    <PO username <PO passwordAPPS username <APPS password

The APPS username and password parameters are not documented and the directory is wrong.

Step 2 of the System Maintenance Upgrade Task (Finishing your upgrade) tells you to run dbcom.drv to install the help files. On Compaq True 64 UNIX we found, that for several modules the help was actually located under $APPL_TOP/gml/help rather then $GML_TOP/help.
The workaround was to move the helpfiles into the correct subdirectory.
It's unclear if this is a platform specific issue. See TAR 1344814.999.

After the 11i upgrade

When completing this upgrade,  you should apply the latest AK cummulative patch. This patch is a prerequisite for ALL patches using AKLOAD.
Currently this is patch 1477394.

Customer having upgraded from previous releases to 11i that have a shared installation of HR need to apply patch 1397089.
If HR is fully installed, then this patch is not needed.

If you have upgraded from 10.7NCA or 11.0.X, then you could very well experience problems in the following areas:

• Accessing help
• Trying to do File->Export from the menu
• Trying to logon to SSWA

ORA-01001: invalid cursor ORA-06512: at "SYS.DBMS_SYS_SQL", line 775
ORA-06512: at "SYS.WPG_DOCLOAD", line 346
ORA-06502: PL/SQL: numeric or value error
ORA-06512: at line 10

If you experience this problem then check out note 115996.1. The solution is to run the following script as SYSTEM to create a drop script to drop the old OWS 3.0.2 packages from the database:

set echo off
set heading off
set pagesize 0
set linesize 110
set feedback off
set term on
spool owa_drop.sql
select   'drop ' || object_type ||' '||owner||'.'||object_name||';'
from dba_objects
where owner not in ('PUBLIC', 'OWAPUB')
and (owner||'.'||object_name not in
('APPS.OWA_INIT','APPS_MRC.OWA_INIT','APPS.OWA_CUSTOM','APPS_MRC.OWA_CUSTOM'))
and (object_name like 'OWA%'  or object_name in ('HTP','HTF','DEBUG_CARTX'))
order by object_type desc, owner, object_name ;
spool off

When this script completes, check the 'owa_drop.sql' script to see if it will actually drop something. If it does, then run the script as SYSTEM.
After this the problems should be solved.

Note: this problem could also exist in the VISION demo database

After upgrading a 10.7 database to R11i you will have a number of invalid objects.
Here are the bugnumbers dealing with these problems:
 

• See bug 1324523 for invalid AMS_* views
• See bug 1324549 for invalid AR_MRC_OSPT_PKG and FA_MRC_OSPT_PKG this issue which is supposed to be fixed in patch 1294694.
• See bug 901221 for invalid PA_PURGE* objects.
• See bug 1331040 for invalid CN_SRP_RULE_UPLIFTS view

R11i NLS/MLS issues

If the database being upgraded has additional languages installed, then you are required to install the NLS version of a the preupgrade patch for each of the languages you have installed. The NLS version of these patches should be installed BEFORE installing the big 11i NLS patch. If you wish you can install the NLS version of the preupgrade patches before running autoupgrade (adaimgr), but it not required as the 11i NLS patch itself is always installed after the upgrade has completed. This automatically implies that autoupgrade (adaimgr) will not do anything with the NLS files on the system.

Regular patches and NLS

If the database you are running has additional languages installed, then you are required to install the NLS version of a patch for each of the languages you have installed. So if the database is running AMERICAN, DUTCH, GERMAN and FRENCH, then after installing the AMERICAN version of the patch, the DUTCH (netdut), GERMAN (gerger) and FRENCH (frafre)  translation of the patch has to be installed as well.

When upgrading from 10.7 or 11.0.X NLS to 11i NLS, there is no need to deactivate and remove the language anymore.
For details check the 11i NLS release notes. It explains what has to be done.

During the NLS installation you could encounter problems which FNDLOAD:

Pid 14376 received a SIGSEGV for stack growth failure.
Possible causes: insufficient memory or swap space, or stack size exceeded maxssiz.

If you get this error, then you've probably run into bug 1400354.
This problem can be fixed by applying patch 1396020.

If you plan to run MLS, then you will have to add a variable to te apps.conf in the <ORACLE_SID>ora/iAS/Apache/Apache/conf/apps.conf file.
To enable MLS add the following line to the end of the script:

• SetEnv NLSSUPPORT=yes

R11i localizations issues

R11i Client issues

If customer is using Internet Explorer to access APPS, then please be aware of bug 1497145. The export functionality will only work after patch 1497145 has been installed.

Issues installing the 11.5.3 maintenance pack

The 11.5.3 patch will modify the appsweb.cfg setting de default connection mode for Oracle*Forms to 'socket'.
If you are currently using 'http' mode, you will need to modify the <COMMON_DIR>/admin/scripts/adfrmctl.sh and change 'http' into 'socket'.

Failing to do so will result into a 'FRM-92060 failed to connect to server' error.

The $FND_TOP/sql/aflobbld.sql script which has to be run as part of the post install steps for 11.5.3 is not rerunnable if it fails. For details see bug 1616283.
If this script fails, the FND_LOBS_CTX index will de dropped. Trying to recreate it means running $FND_TOP/patch/115/sql/AFLOBCX3.sql. This script in turn is likely to fail with the following error:

ERROR at line 1:
ORA-29855: error occurred in the execution of ODCIINDEXCREATE routine
ORA-20000: interMedia Text error:
DRG-50857: oracle error in dreiiflush
ORA-1631: max # extents (505) reached in table APPLSYS.DR$FND_LOBS_CTX$I
ORA-6512: at line 46

sqlplus system/manager
sql> alter tablespace APPLSYSD default storage (PCTINCREASE 50);
sql> alter tablespace APPLSYSX default storage (PCTINCREASE 50);
sql> connect  apps/apps
sql> @$FND_TOP/patch/115/sql/AFLOBCX3.sql applsys apps
sqlplus system/manager
sql> alter tablespace APPLSYSD default storage (PCTINCREASE 0);
sql> alter tablespace APPLSYSX default storage (PCTINCREASE 0);

While installing prerequisite patch 1526604 on WINDOWS NT, we found that it complained about missing driverfiles pvfile.drv and aslfile.drv.
These driver files will actually be installed by the 11.5.3 patch. As a workaround we created dummy 0 byte files which allowed the patch to continue.

NLS customer applying the 11.5.3 maintenance pack should install patch 1625993 to prevent the following error when running pqcptca.ldt:

Uploading PQH_TRANSACTION_CATEGORIES POSITION_TRANSACTION
A database error occurred:
  ORA-6550: line 9, column 1:
PLS-306: wrong number or types of arguments in call to 'LOAD_ROW'
ORA-6550: line 9, column 1:
PL/SQL: Statement ignored

Apps related RDBMS issues

If you are using RDBMS version 8.1.6.3, then there are some issues that you have to be aware of.

First of all, you must install the backport for patch 1640583 when running 8.1.6.3. This is required as part of the APPS certification.
The backport numbers are:
 

Platform	Backportnumber
IBM RS6000	1645007
HP-UX	1644999
Compaq True 64 Unix	1645009
Sun Solaris	1644994
Windows NT	1645013

To prevent 'ORA-4030 Deadlock detected' during package compilation, set the following parameter in the init.ora:

aq_tm_processes=0

To prevent 'ORA-3113: end-of-file on communication channel' while compiling PO_WF_NOTIFICATIONS_V or while selecting complex views, set the following parameter in the init.ora:

_push_join_predicate=FALSE

Related to the above ORA-3113 seems to be bug 1584713 which fixes this error in the Segment Flexfields screen. It's unclear if this is the exact same problem. For APPS patch 1584713 has been release which fixes this specific issue (but not the PO_WF_NOTIFICATIONS_V) issues.

Customers running Order Management can run into the following error in the Enter Sales Order screen after upgrading to 8.1.6.3:

FND_AS_UNEXPECTED_ERROR
(PKG_NAME=ONT_HEADER_Def_Hdlr)
(PROCEDURE_NAME=Default_Record)
(ERROR_TEXT=OFND_AS_UNEXPECTED_ERROR)

Also, to fix bug 1034486, the following event must be added to the init.ora:

event = "10943 trace name context forever, level 2"

Port specific problems

Sun Solaris

R11i installation on Sun Solaris will only succeed if you are running Solaris patchlevel 20 or higher (patch 105284-20 or above). See also bug 1312252.

HP-UX

In order to get admrgpch working on HP, you need to apply the following patches (in that order):

• 1430909 (AD) or the latest AD patch, currently patch 1553656
• 1405438 (fix for affile.lc)

On HP you might run into bug 1340492. This means, that when the rapidwiz is relinking oracle it fails with the following error:

pxdb32: warning: corrupted PD table
pxdb32: warning: corrupted PD table
pxdb32: internal error. File won't be debuggable (still a valid
executable)
*** Error exit code 10

This will allow you to continue the installation.

According to bug 1340492 this problem can be solved by applying HP patches:

PHSS_21223 (compiler cumulative patch)
PHSS_19866 (linker cumulative patch)

If you plan to use CRM on HP-UX, make sure you have applied intermedia patch 1344060. If not you will run into the following error when applying the d1306413.drv on file cskbctxd.sql :
 

AD Worker error:
The following ORACLE error:

ORA-29856: error occurred in the execution of ODCIINDXDROP routine
ORA-20000: interMedia Text error:
ORA-1722: invalid number
ORA-6512: at "CTXSYS.DRUE", line 126
ORA-6512: at "CTXSYS.TEXTINDEXMET

occurred while executing the SQL statement:
DROP index cs_kb_sets_tl_N1

IBM RS6000

If you accidentally start the rapidwiz in a UNIX session that has the JAVA_HOME variable set, then you will encounter the following error:

Can't find class java.lang.System
Could not create Java VM

Use SMIT to verify that you have installed the fileset called X11.samples.lib.Core. If it is missing install it and add the following
path to your LIBPATH:

/usr/lpp/X11/Xamples/lib/Xmu

This fileset will deliver the libXmu.a which is needed if you want to relink f60webmx.

If you run rapidwiz as root, make sure that the 'umask' is set to 022. If it's set to 077 or 027, then the Oracle Universal installer will start asking all the questions that you basically answered using the rapidwiz. This is fixed in 11.5.3.

AIX customers should install iAS patch 1528039. This will fix a core dump in Apache which happenen when you were trying to logon via the Personal Homepage. Without this patch, you cannot logon via the Personal Homepage or use the online help.

In order to be able to use the 8.1.6 loadjava command, you need RDBMS patch 1309395.

Windows NT

In order to succesfully install windows NT, make sure that the c:\winn\system32\driver\etc\hosts file contains the IP address and the fully qualified host name (hostname + domain). If this is not the case, then the rapidwiz will not continue to the next screen at certain points (depending on the type of installation).
Also make sure, that the NT box has a fixed IP address assigned and that it does not get the IP address from a DHCP server.

When installing the MKS and the Visual C++, you should ensure that the directory in which these products are installed does not contain any blank spaces. This is because otherwise relinking will fail. Therefor the folder 'Program Files' is not the directory in which you want to install.

While installing the GNU Make utilitiy we found, that the only version which seems to work is the GNU make 3.79.1 version.

Manually relinking executable on NT doesn't work in a DOS box after running APPSORA.cmd. Adadmin will allow you run relink files, but it will not allow you to relink AD executables. These should still be manually relinked. For example:
    sh adrelink.sh force=y ranlib=y "ad adpatch.exe"

Bug 1397908 has been raised for this issue. The release notes actually tell you how to manually relink on NT, but it's different from UNIX.
To manually relink on NT, you have to create a APPS.sh using adregenv.exe:
    C:\> adregenv.exe <APPL_CONFIG>

This will create the APPS.sh. When completed you can manually relink AD using the following steps:

• Open a DOS box
• Go to the %APPL_TOP% directory
• Type 'sh' to open a shell
• Type '. ./apps.sh' to source the environment file
• Type e.g. 'sh adrelink.sh force=y "ad all"' to relink all AD executables.

It appears that adjri.zip version 115.11 and 115.13 from 1394852 and 1399833 and 1420362 is causing problems on NT. Applying one of these patches will cause adjava  to hang when generating JAR files. The solution is to install patch 1411929.
This problems is fixed when applying AD patch 1460350 or the latest AD high-priority upgrade patch, currently 1553656

When completing an install on NT, be sure to change the CLASSPATH setting in the following files:

• %APPL_TOP%\admin\adovars.env
• %APPL_TOP%\admin\adovars.cmd

When doing an upgrade on windows NT, the 8.1.6 listener service will not be created.
To create it manually run:
    <common dir>\admin\install\adsvdlsn.cmd

For details see bug 1320811. Development is assuming that if the rapidwiz is not creating a database,  that the database and the listener will already exist or will be build manually.

If you want to be able to merge patches on windows NT using admgrpch.exe, you have to apply the following patches:

• 1430909 (AD) or the latest AD patch, currently patch 1553656
• 1425282

After installing 11.5.2 on NT, you will probably find that HELP isn't working. For details, please check bug 1482758.
The workaround is to delete the "wv_gateway_cfg" system environment variable from the Control Panel/System and then reboot the machine.
This will be fixed in 11.5.3.

Afer completing an 11.5.2 installation on NT, you will find that scripts like 'loadjava.bat' in the 8.1.6 ORACLE_HOME still contains variables like %s_jreLocation%. This problem has been reported in bug 1397424. It has been fixed in a later version of the Universal installer, but this one is not delivered with APPS. Bug 1532929 has been raised for APPS refering to bug 1397424 . Bug 1532929 has been updating requesting development to provide correct values for the variables which have not be replaced. As a result of this problem you will not be able to run loadjava for the xmlparser.jar or any other jarfile untill you replace the variables by correct values.

After doing an 11.5.3 installation on NT, you will probably find that none of the JTP .jsp files are working (jtflogin.jsp, jtfdefaultlogin.jsp etc.).
The reason for this is that the vitual PATH for the /html/ is missing in '/' at the end.

The workaround is to manually modify the <ORACLE_SID>ora\iAS\Apache\Apache\conf\apps.conf and change the alias for the /html/ virtual PATH. A correct example setting is:

Alias /html/ "d:\oracle\v11icomn\html/"
<Location /html/>
  Order allow,deny
  Allow from all
</Location>

Issues installing platinum V3.1

The platinum V3.1 is based on a cloning scenarion (it will replace both database and APPL_TOP). It does however not use the official cloning white paper. This issue has been raised in bug 1671506.

Bug 1663140 was raised as recreating the controlfile fails if the database is named something else then CRMPLAT1.
The workaround is to modify the .sql scripts and change the following:

CREATE CONTROLFILE REUSE SET DATABASE CRMPLAT1 RESETLOGS NOARCHIVELOG

CREATE CONTROLFILE REUSE SET DATABASE <your database name> RESETLOGS NOARCHIVELOG

• crctl_tar.sql
• crctl_vis_tar.sql
• crctl_vis_SID.sql

When preserving the file as described in the 'Required Post Installation steps' you have to store ALL files in the same directory. You should not create any subdirectories in the directory where you are preserving the files. For details see bug 1671569.

Bug 1671496 was raised for the fact that the $OA_HTML/env.txt file is not modified by the plat31_sun_apps_install script. The workaround is to update this file manually and make sure it's referencing the correct APPL_TOP.

Once the installation is complete, you have to manually run the following profile scripts to update some profiles:

cd $<COMMON_TOP>/admin/install
sqlplus apps/apps @adcmprf.sql
sqlplus apps/apps @adfrmprf.sql
sqlplus apps/apps @adwebprf.sql

When using the platinum V3.1 software, then be aware of the fact that it does not include the high priority patches from development.
For details see bug 1671587.

Running 10.7 or 11.0.X against 8.1.6

When running 10.7 against RDBMS 8.1.6, you will need:

• Patch 692543 to fix ARP_AA_HISTORY_PKG package
• Patch 623359 to fix INVWATPB package
• Patch 1332319 (one off for 1054692) to fix ARP_AA_HISTORY_PKG, ARP_CR_ICR_LINES_PKG and ARP_MISC_CASH_DIST_PKG
• Patch 1344223 to fix PO_ASL_AUTHORIZATIONS_THS package (use workaround documented in this bug)
• Patch 1641047 to fix AP_INV_SELECTION_CRITERIA_PKG package.

• Patch 1054683 to fix ARP_AA_HISTORY_PKG, ARP_CR_ICR_LINES_PKG and ARP_MISC_CASH_DIST_PKG
• Patch 1238630 to fix PO_ASL_AUTHORIZATIONS_THS package

•      AP Patch Set Q (or higher)
•      AR Patch Set P (or higher)
•      AS Patch Set H (or higher)
•      AX Patch Set F (or higher)
•      OE Patch Set G (or higher)
•      PO Patch Set M (or higher)