Release 9.0.3.0.0 for Linux
Part Number B10231-01
 Contents |
| Oracle Collaboration Suite Release Notes Release 9.0.3.0.0 for Linux Part Number B10231-01 |
|
This chapter summarizes component issues associated with Oracle Collaboration Suite. Topics include:
In the Thin Client, messages forwarded as an attachment do not work by default.
Workaround: Users must select the "as attachment" option:
The following issues are associated with Oracle Wireless:
Oracle Enterprise Manager displays all of the processes that can be managed for a middle-tier machine. When accessing the EM page and clicking on a middle-tier machine, Wireless will show a red down arrow. The reason for this is that the Wireless Server has not been started. To start the Wireless Server, from your browser, point to:
@ http://<midtier_machine_name>:<midtier_webcache_port>/ptg/rm
This will automatically start the Wireless Server.
You can receive a wireless notification for a message received by Oracle Email. If the ID of either the subject or the sender in the original message contains multi-byte characters, then these multi-byte characters are not displayed correctly in the notification.
Oracle Collaboration Suite native characters are displayed as question marks in Oracle Files. Folder names created with non-ASCII characters are correctly displayed, but end-users are not able to navigate into these folders.
Help information for Oracle Wireless Modules for Files, Directory and Address Book is displayed correctly only if the users' language is set to English.
Users cannot create Calendar entries using the Oracle Wireless Calendar Module if the entries contain non-ASCII characters. This may be the case if users' language is set to a language other than English.
If both the infrastructure and middle tiers are installed on the same machine, and if the Enterprise Manager daemon running on that computer is referring to the infrastructure home, then the "Register Oracle Portal Provider for Wireless Webtool" and "Register Oracle Portal Provider for Wireless Customization" from the Wireless site will create "java.lang.NoClassDefFoundError" errors.
In the orion-web.xml file of the infrastructure installation, pdkjava.jar and ptlshare.jar should refer the middle tier installation location.
For example, if the infrastructure is installed in:/private/ias20_infra/
and the middle tier is installed in:/private/ias20_midtier/
then the following entries in the Enterprise Manager daemon orion-web.xml:
<classpath path="/private/ias20_infra/portal/jlib/pdkjava.jar"/><classpath path="/private/ias20_infra/portal/jlib/ptlshare.jar"/>
should be replaced with:
<classpath path="/private/ias20_midtier/portal/jlib/pdkjava.jar"/><classpath path="/private/ias20_midtier/portal/jlib/ptlshare.jar"/>
Restart the opmn process for the changes to take effect.
A generic error affects some Oracle9iAS components, including Oracle Wireless. This errors occur when you use Microsoft Internet Explorer to access the Web tool on a machine that has both infrastructure and a middle tier installed on it. You may encounter the following errors:
SimpleImage supports an alt attribute which provides alternate text for display devices, and an alternate .wav file for the voice deviceclass devices. For example:
<SimpleBind><SimpleMatch> </SimpleMatch> <SimpleTask> </SimpleTask> <SimpleDisplay><SimpleTextItem deviceclass="voice"><SimpleAudio src="http://somehost/audio.wav"/></SimpleTextItem> <SimpleTextItem deviceclass="pda, microbrowser">Hello welcome</SimpleTextItem></SimpleDisplay></SimpleBind>
The Modules included in this production release are complete and functional. There are, however, important updates to the modules that will soon be released as a patch.
You can use the Modules as they are, but be advised that the patch will make significant database and other changes, requiring you to discard previous end-user personalization information. You will not be able to migrate your previous information when you install the patch.
The Web Component Developer is a java-based GUI tool that allows you to create Web Components that can be delivered by the Web Component Provider, as Wireless Services. There is a known issue of the Web Component Developer in the generation of Web Component Definition files. This renders the developing of composite Web Components not possible. This known bug is scheduled to be fixed and delivered to customers via the Automated Release Updates.
Oracle Wireless applications cannot be clustered using the Oracle9iAS clustering mechanism. However, it is still possible to configure Oracle9iAS such that a high availability deployment can be achieved. Complete the following steps for high availability deployment.
ORACLE_HOME/dcm/bin/dcmctl stop
ORACLE_HOME/bin/emctl stop
ORACLE_HOME/opmn/conf/ons.conf on each of the middle tiers and on the infrastructure contains IP address entries for all the other tiers. If not, file and add missing IP-address entries.
OC4J_Wireless OC4J instance to the desired number. This can be done from the EM console or by modifying the file ORACLE_HOME/opmn/conf/opmn.xml.
For instance, if you modify opmn.xml, then a typical entry to start four OC4J processes in the default island would be of the form:
<oc4j maxRetry="3" instanceName="OC4J_Wireless" gid="OC4J_Wireless" numProcs="4">
mod_oc4j configuration file for each middle tier (that is: ORACLE_HOME/Apache/Apache/conf/mod_oc4j.conf), modify the mount point entries for the Wireless runtime. If two middle tiers (M1 and M2) are used, then the entries should be of the form:
Oc4jMount /ptg instance://m1.c1.mysite.com:OC4J_Wireless,m2.c2.se4637-u-sr006.us.oracle.com :OC4J_Wireless
and
Oc4jMount /ptg/* instance://m1.c1.mysite.com:OC4J_Wireless,m2.c2.se4637-u-sr006.us.oracle.com :OC4J_Wireless
where c1 and c2 are the respective Oracle9iAS instance names and can be determined by running the command:
ORACLE_HOME/dcm/bin/dcmctl whichInstance
These entries should be exactly the same for all middle tier machines.
ORACLE_HOME/dcm/bin/dcmctl updateConfig to update the DCM repository with the configuration file changes.
On slow machines, it is likely that a DCM error (timeout) of the form ADMN-906005 is displayed. If this occurs, run the command ORACLE_HOME/dcm/bin/dcmctl getReturnStatus and wait until the command exits. This confirms that the changes have been propagated to the DCM repository.
<cluster-config/> under the <orion-web-app> tag in the following file:
ORACLE_HOME/j2ee/OC4J_wireless/application-deployments/ptg/ptg-web/orion-web .xml.
ORACLE_HOME/dcm/bin/dcmctl start
ORACLE_HOME/bin/emctl start
Currently, high availability support is only available for the core server runtime (by default mapped to the URI /ptg/rm).
Oracle Wireless process status is unavailable for multiple ORACLE_HOMEs with external repository.
When installing both middle and infrastructure tiers on the same machine and changing the Wireless schema from the Enterprise Manager console to point to a schema other than the one available as part of the infrastructure installation, the Wireless process status changes are not displayed on the Enterprise Manager console. This problem occurs on all platforms.
Here is the workaround for this problem:
From the ORACLE_HOME of the middle tier for which the schema has been changed, copy the following fragment from the file
<middle-tier ORACLE_HOME>/config/iasschema.xml
and paste it over (overwrite) the corresponding entry in the infrastructure ORACLE_HOME file
<infrastructure ORACLE_HOME>/config/iasschema.xml <SchemaConfigData> <ComponentName>Wireless</ComponentName> <BaseName>WIRELESS</BaseName> <Override>true</Override> <SchemaName>the new schema name</SchemaName> <DBConnect>the new DB connect string</DBConnect> <Password>the new DB password (encrypted)</Password> </SchemaConfigData>
Restart Enterprise Manager after this is done.
The following issues are associated with Oracle Files:
In Oracle Files CA, if the "Integrate with Workflow" Option is checked during the configuration of Oracle Files, then the configuration will fail while uploading the translated workflow process definitions.
Action: Contact your Oracle Support Representative for the manual steps required to successfully complete the configuration of Oracle Files integrated with Workflow.
From either the Oracle Files portlet or the Oracle Collaboration Suite Search portlet, the end-user cannot search for files or folders with multi-byte characters in their names.
Action: The end-user can navigate directly to the Oracle Files application or the Oracle Collaboration Suite Search application to execute their search successfully.
If files contain multi-byte characters in their filenames, then the end-user cannot upload such files into Oracle Files through the Netscape 4.7x browser interface.
Action: Use Microsoft Internet Explorer 5.x and above.
The end-user cannot enter multi-byte characters in the folder name field when trying to create a new folder in Oracle Files.
Action: Use Microsoft Internet Explorer 5.x and above.
Any end-user that has a username that contains both multi-byte characters and underscores ('_') cannot login to Oracle Files.
Action: If usernames must contain multi-byte characters, then do not also include underscores in these usernames.
After installing and configuring Oracle Files, there is a possibility that the Oracle Files target are missing, because the targets.xml file does not include the target in the Oracle Enterprise Manager interfaces. When logging in to the Oracle Enterprise Manager Web site using host name:1810, the Oracle Files targets might be missing.
Action: Stop Oracle Enterprise Manager and re-run the Oracle Files Configuration assistant.
emctl stop
$ORACLE_HOME/ifs/files/bin/ifsca
emctl start
The following issues are associated with Oracle Calendar Web Client:
If Oracle Calendar web client Accessible mode is disabled after installation, then the ADA parameters in $ORACLE_HOME/Apache/Apache/must be modified as follows
fcgi-bin/lexacal-private/ini/webcal.ini :
hide_toggle_link
If the g: or r: prefixes are used when performing a search from the Edit Access Rights page, then the following error messages appear and are misleading:
The error message should say Invalid search criteria.
When 12:00 p.m. is selected from the drop-down box on the Suggest Time page, 12:00 a.m. is set in the General section of the Entry Modification/Creation pages. This is only an issue when you are using the 12-hour time format for the web client.
Workaround: Set the 24-hour time format as your preference.
When clicking Suggest Time from the Create/Modify Meeting page, the Suggest Time dialog box opens but appears behind the main page.
Workaround: Resize the web client main page and click the Suggest Time dialog box to bring it to the foreground.
When setting values for the Print section of the Preferences page, using a value greater than 3 for the Margin setting causes the web client to freeze.
You can only add an existing Group to a new Group by:
Or
Netscape 4.x on Linux is not supported by the Oracle Calendar web client.
The following issues are associated with Oracle Calendar Server:
Portions of the IMAP4, SMTP, and MIME implementations are based on the Netscape Access Messaging Software Development Kit (SDK) code. The Original Code is Mozilla Communicator client code, released March 31, 1998. The initial developer of the original code is Netscape Communications Corporation.
Portions created by Netscape are Copyright ©1998-1999 Netscape Communications Corporation. All rights reserved.
Changes made to the Netscape Messaging Access SDK by Oracle are located at:
http://otn.oracle.com
The source code containing changes to the Netscape Access SDK by Oracle can be downloaded in ZIP format at:
http://otn.oracle.com
The first time you sign in to the Calendar Administrator, some browsers may encounter a Page Not Found error. If this happens, then use your browser Back button to return to the Calendar Administrator sign-in page, and sign in again.
If you experience a problem with incorrect timestamps on calendar notifications, set the /users/unison/misc/unison.ini [CWS] mailhdrdateusetimezone parameter to TRUE on the Oracle Calendar server host.
The Calendar Administrator is installed by default as a CGI application, not a FastCGI. If you need to set the path to the Calendar Administrator manually in the ThinClientNavigationBar.properties file as instructed in the "Postinstallation Tasks" section of the Oracle Collaboration Suite Administrator's Guide, then the URL to use should be:
https://<host>:<port>/cgi-bin/uniwebadm.cgi?CADMType=CADMClusterAdmin
The CAPI_AuthenticateAsSysop function is not supported in this release.
When used with Oracle Email, Oracle Outlook Connector does not support the following:
Note that the Oracle Thin Client interface can grant access to mail folders. These folders may then be opened using Oracle Outlook Connector.
The following issues are associated with Oracle Ultra Search:
During the ETL (Extract-Transform-Load) migration from Oracle Ultra Search version 1.0.3 to 9.0.3, the wk0migrate.sql script produces the following errors and then the migration and fails.
ERROR: failed to migrate query statistics for instance "<instance_name>" ORA-00942: table or view does not exist ... migrating URL table ... start migrating URLs for instance "<instance_name>" ERROR: failed to migrate URL table for instance "<instance_name>" ORA-00942: table or view does not exist
Workaround: Update the wk0migrate.sql script prior to running it
PROMPT ... creating database views based on source database instance tables declare l_stmt varchar2(4000); l_schema varchar2(100); begin for inst in (select inst_schema from wk$instance_tmp) loop l_schema := inst.inst_schema; -- create view wk$url_tmp l_stmt := 'create view '||l_schema||'.wk$url_tmp as select * from ' || l_schema||'.wk$url@src_db '; execute immediate l_stmt; -- create view wk$statistic_tmp l_stmt := 'create view '||l_schema||'.wk$statistic_tmp as select * from ' || l_schema||'.wk$statistic@src_db '; execute immediate l_stmt; end loop; end; /
PROMPT ... re-creating all of the user instances exec wk_upgrade.recreate_instances;
Paste the above section here
PROMPT ... migrating system administration configuration exec wk_upgrade.migrate_data('SYS_ADMIN');
When using OPCA (Oracle Portal Configuration Assistant) to migrate Oracle Ultra Search from version 1.0.3 to 9.0.3, users see numerous errors in loading Portal database objects. As a result, Oracle Ultra Search migration does not go through.
Workaround: Bypass the steps of using OPCA and follow the manual migration steps described in the section "Ultra Search Migration Approaches" of the Oracle Ultra Search User's Guide.
The Ultra Search remote crawler does not run, and no crawler log file is produced. In the Ultra Search Administration Tool, an error message for the schedule displays the following exception:
Remote crawler launcher encountered one or more problems while attempting to locate remote object on host: localhost java.rmi.activation.ActivateFailedException: failed to activate object; nested exception is:
java.rmi.activation.ActivationException: exception in object constructor; nested exception is:
java.util.MissingResourceException: Cannot find bundle for base name oracle.context.isearch.IsearchResBun, locale en_US
To expand the Ultra Search EAR files, start up OC4J successfully at least once on the OCS Middle-tier installation. This will expand the Ultra Search EAR files.
$ORACLE_HOME/j2ee/OC4J_Portal/config/server.xml
There is a web application named UltrasearchAdmin. Find the path to its EAR file. Locate the directory where the EAR file is expanded into. This is contained in the same directory where the EAR file is located. The expanded directory name should be the same as the EAR file name (without the .ear extension). This directory is referred to through the EXPANDED_ADMIN_DIR variable.
isearch_midtier.jar file.
isearch_midtier.jar directly in $EXPANDED_ADMIN_DIR.
isearch_midtier.jar to the ultrasearch/lib directory on the remote crawler host.
$ORACLE_HOME/ultrasearch/lib.
isearch_midtier.jar into this directory.
If the RMI registry and daemon are already running, then you must ensure that the isearch_midtier.jar file is picked up by restarting both these processes.
The bug is due to the IsearchResBun resource bundles not found by remote crawler launching system. The resource bundles are contained in isearch_midtier.jar.
isearch_midtier.jar is not shipped in the ultrasearch/lib directory in this release. As a result, a java.rmi.activation.ActivateFailedException is created, which is fundamentally triggered by a java.util.MissingResourceException.
The only place where isearch_midtier.jar can be found is in the Ultra Search Administration Tool EAR file located in the OCS Middle-tier installation Oracle Home. The solution involves copying the isearch_midtier.jar file from the OCS Middle-tier installation to the ultrasearch/lib directory in the remote crawler host.
Oracle Ultra Search schedule might fail in an Ultra Search instance created as a SSO user.
To workaround this issue, grant the database user the Ultra Search super user privilege ('wkproxy'). To do this, complete one of the following:
SQLPLUS as "wksys" or "sys", and type the following:
exec wk_adm.grant_sysadmin('WKPROXY')
or,
orcladmin", "Portal", or "wksys". Then go to the "Users" tab and "Super-Users" subtab. Select user type as "Database User", and the user name as "wkproxy", and click Add.
Perform the above mentioned workaround before a SSO user creates Ultra Search instances. Otherwise, the schedule will fail for the Ultra Search instance created by a SSO user.
Note that Ultra Search instances created by a database user will not experience the schedule failure problem. A database user can go to
http://<hostname>:<port>/ultrasearch/admin/index.jsp
where hostname:port should point to your Oracle9iAS instance to create an Ultra Search instance.
Ultra Search provides the SQL script "wk0upgrade.sql" for migrating the user data and database objects from an existing Ultra Search 9.0.1 database to a migrated Ultra Search 9iR2 database. However, users will generate errors when running into the stage "stop all of the crawler schedules and database jobs". Also, the script may fail and stop during the stage "re-creating all of the user instances". You can request a patch to fix this problem.
On RedHat Linux, the INSO filters may report a document conversion failure when the Ultrasearch crawler fetches documents of certain types.
$ORACLE_HOME/ctx/bin/ctxhx: error while loading shared libraries: $ORACLE_HOME/ctx/lib/libsc_ut.so: undefined symbol: stat
To workaround the issue on a 9.0.1.x database, apply the patch for bug 2037255. To workaround the issue on a 9.2.0 database, apply the patch for bug 2446892.
The following known issues are associated with Oracle HTTP Server:
If you are running more than 1024 Oracle HTTP Server processes in the same box that are accessing OC4J instances, then set the Oc4jCacheSize directive in mod_oc4j.conf to 0 (zero). This disables the use of persistent connections between mod_oc4j and OC4J instances and improves performance.
After installation of Oracle9iAS, the following error messages may appear in the error_log file in $ORACLE_HOME/Apache/Apache/logs:
[Fri Apr 5 15:17:50 2002] [error] (2)No such file or directory: FastCGI: access for server (uid -1, gid 10) failed: read not allowed by group [Fri Apr 5 15:17:50 2002] [error] (2)No such file or directory: FastCGI: can't create dynamic directory "/tmp/fcgi_10258/dynamic": access for server (uid -1, gid 10) failed: read not allowed by group
These error messages are harmless and will not affect the operation of FastCGI. They are caused by a re-ordering of directives in httpd.conf during installation. To eliminate these messages, the directives User and Group must be moved from their location at the end of the file to a location before the FastCGI server directives in httpd.conf.
If an infrastructure Single Sign-On Server installation and a middle tier installation are on the same machine (in different Oracle homes), then Microsoft Internet Explorer reports an incorrect host header after a redirect. This incorrect host headed causes mod_osso to generate an error message when trying to access a protected resource after the user has been redirected from the Single Sign-On Server back to the original server. If you click on Reload in Internet Explorer, then the session continues successfully. This issue will not occur if any of the following conditions are true:
When you install an infrastructure instance of Oracle9iAS and a middle tier installation on the same machine, Microsoft Internet Explorer might report various errors where an incorrect host header is sometimes passed after redirection. Specifically, if you have already logged on via the Single Sign-On Server to the middle-tier instance, and if you subsequently click on a link that tries to redirect them to the infrastructure instance, then you will receive an OSSO error page. Pressing the Back button allows you to continue to the desired page.
If you issue the "opmnctl stopproc" command in a process within seconds after the process was killed or abnormally terminated, then the "opmnctl stopproc" command might hang. This might prevent you from issuing other process-related commands.
In this situation, issue the following commands:
opmnctl reload opmnctl stopproc
If you are using dcmctl, then the "dcmctl stop" command might not hang, but fail instead. Issue the following commands to resolve the situation:
dcmctl updateconfig opmn dcmctl stop
Perform the following steps to protect metadata directories that are used by OJSP:
httpd.conf file:
<DirectoryMatch /WEB-INF/> Order deny,allow Deny from all </DirectoryMatch>
ojsp.conf file by changing:
<Location /_pages/> Order deny,allow Deny from all </Location>
to be:
<DirectoryMatch /_pages/> Order deny,allow Deny from all </DirectoryMatch>
After installation, if you wish to configure Virtual Hosts in the httpd.conf file by using the 'advanced' section of the EMD console or by using a text editor on the httpd.conf file, then use the following guidelines:
VirtualHosts are provided after the Port, Listen, and ServerName directives. A simple example of a correctly set Virtual Host section might be as follows:
# # these are set at the end of the httpd.conf file after the IAS installation occurs # Port 7778 ServerName someServer.mycompany.com Listen 7779 # # these lines were added manually to create a virtualHost # NameVirtualHost 1.2.3.4 <VirtualHost 1.2.3.4> DocumentRoot /u01/app/oracle/product/iAS9020_portal/Vhost1.htdocs ServerName Vhost1.mycompany.com </VirtualHost>
dcmctl utility to update your changes.
<ORACLE_HOME>/dcm/bin/dcmctl updateConfig
This is automatically done for you if you edit the file through the EMD console.
The following known issues are associated with Oracle9i XML Developer Kit:
The following is additional support provided by Oracle9i XML Developer Kits (XDKs):
This following lists additional support and options provided by Oracle9i XML Developer Kits (XDKs):
XSLT Processor for Java is now threadsafe.
The following known issues are associated with Oracle9iAS Portal:
When two Oracle9iAS Oracle HTTP Servers are running on a single machine, such as when an infrastructure installation and a Portal and Wireless middle-tier installation are both installed on a single machine, then a login to the Portal is not possible when using a Microsoft Internet Explorer (MSIE) Browser. This has been verified with versions 5.5 and 6.0 of MSIE. The problem occurs when a browser redirect is issued from one port used by the Oracle9iAS Single Sign-On server to the port used by Oracle9iAS Web Cache, which is fronting Oracle9iAS Portal. When MSIE receives the redirect, it erroneously sets the Host: header with the first port rather than the destination port. This behavior is not exhibited by Netscape Navigator (versions 4.6 and 4.7). When this error occurs with MSIE, one of the following messages may be shown after an attempt to log in:
Error: Unexpected error encountered in wwsec_app_priv.process_signon (ORA-06502: PL/SQL: numeric or value error: character string buffer too small) (WWC-41417)
or
Error: The decryption of the authentication information was unsuccessful. This may be caused by corruption of the data, an incorrect encryption key in this application's configuration, or an illegal access attempt. Please notify your administrator. (WWC-41454)
Although MSIE sets the Host: header incorrectly in either SSL or non-SSL mode, the problem shows up only in SSL mode because Oracle9iAS Web Cache is set up to map ports to a catch-all default HTTP port in non-SSL mode. The workaround is to use two separate machines with the two installation types.
You cannot create PL/SQL functions and procedures in languages other than English when using Oracle9iAS Portal from a browser. You must create them using SQL*Plus.
The link to Portal Monitoring Services does not work, this is the link displayed on the Administer tab.
A workaround is to cut and paste the following block of code into a SQLPLUS session that is connected as sys or portal on the same database as the PORTAL schema. If the portal schema name is not PORTAL, then the script will need updating accordingly in the two places that are indicated by the term "PORTAL" highlighted in bold in the following code example:
/* * * This patch script inspects and updates the Portal Service Monitoring * link (to the Oracle Enterprise Manager Portal Target) on the * Portal Administer Tab. It strips the link of redundant extra host * entries, if necessary i.e. the URL on 9.0.2 incorrectly has the * following parameters: * $target=<iASName>.<host>%5F<host>%5FPortal%3Aportal%3A7778$ctxName1=<iASName>.<host>%5F<ho st> * it should be of the form: * $target=<iASName>.<host>%5FPortal%3Aportal%3A7778$ctxName1=<iASName>.<host> * * There is logic in the script to prevent it running twice. * NOTE: if running against a PORTAL schema named something other than * PORTAL please update the two PORTAL schema references accordingly */ DECLARE CURSOR c1 is SELECT url FROM PORTAL.wwptl_other_services_link$ where NAME='MONITORING'; workerURL VARCHAR2(2000); newURL VARCHAR2(2000); host VARCHAR2(1000); testIndex NUMBER; indexOfStartClip NUMBER; indexOfEndClip NUMBER; urlLength NUMBER; BEGIN OPEN c1; FETCH c1 INTO workerURL; indexOfEndClip:= INSTR (workerURL,'%5FPortal%3',1,1) ; urlLength := LENGTH(workerURL); newURL := SUBSTR (workerURL, 0, indexOfEndClip-1); indexOfStartClip:= INSTR (newURL,'%5F',-1,1) ; /********************************* * Test to see if the patch has already been run * and if so don't do anything * if the ias$ string is present the extra host has * already been clipped *********************************/ host := SUBSTR (newURL, indexOfStartClip+3, indexOfEndClip); testIndex := INSTR (host,'ias$',1,1); IF testIndex = 0 THEN newURL := SUBSTR (newURL, 0, indexOfStartClip); workerURL := SUBSTR(workerURL, indexOfEndClip+1,urlLength-indexOfEndClip); indexOfStartClip:= INSTR (workerURL,'%5F',-1,1) ; workerURL := SUBSTR(workerURL, 0,indexOfStartClip-1); newURL := newURL||workerURL||'?event=doLoad'; update PORTAL.wwptl_other_services_link$ set URL = newURL where NAME = 'MONITORING'; COMMIT; END IF; CLOSE c1; END; /
Next you need to clear the mod_plsql cache and restart Oracle9iAS Web Cache to flush out existing pages with the incorrect link. To do this:
plsql directory:
rm -fr ORACLE_HOME/Apache/modplqsl/cache/plsql/*
cd ORACLE_HOME/webcache/bin ./webcachectl restart
NLS cannot be authenticated by Oracle9iAS Single Sign-On if the username contains non-ASCII characters.
For non-ASCII usernames to work, the default encoding of the OC4J Java VM that is running the Oracle9iAS Syndication Server should be compatible with the character set of the Oracle9iAS Single Sign-On Login Server. The Java VM default encoding is determined by the locale on which OC4J is started. The character set of the Oracle9iAS Single Sign-On Login Server is the character set of the database on which the Oracle9iAS Single Sign-On Login Server is deployed.
For example, when OC4J is deployed on a Japanese locale environment with Shift-JIS as the character set of the locale, the default encoding of the Java VM is SJIS (Java encoding name for the Shift-JIS character set). In this case, the database character set of the Oracle9iAS Single Sign-On Login Server should be JA16SJIS (the Oracle character set for SJIS) in order to use Japanese multibyte usernames.
If you do not configure Oracle9iAS Portal during installation time, and if you want to configure at a later time through Oracle Enterprise Manager, then the portal configuration might fail with the following error:
"The component failed during the configuration operation. Correct any errors reported on this page and in the error logs and retry the operation."
The workaround is to manually configure Oracle9iAS Portal during post installation by performing the following steps:
ptlem.sh script from ORACLE_HOME/assistants/opca directory.
If both the infrastructure and middle-tier are installed in the same machine, and if the EMD running in the machine refers to the infrastructure Oracle home, then the "Register Oracle Portal Provider for Wireless Web tool" and "Register Oracle Portal Provider for Wireless Customization" from the Wireless site display "java.lang.NoClassDefFoundError" errors.
As a workaround, be sure that the Oracle9iAS Portal jars, pdk java.jar and ptlshare.jar, in the orion-web.xml of the infrastructure installation both point to the middle-tier installation.
For example, if the infrastructure is installed in location /private/ias20_infra/, and if the middle tier is installed in location /private/ias20_midtier/, then the following entries in the EMD's orion-web.xml:
<classpath path="/private/ias20_infra/portal/jlib/pdkjava.jar"/> <classpath path="/private/ias20_infra/portal/jlib/ptlshare.jar"/>
should be replaced with
<classpath path="/private/ias20_midtier/portal/jlib/pdkjava.jar"/> <classpath path="/private/ias20_midtier/portal/jlib/ptlshare.jar"/>
Please restart the opmn process for the changes to take effect.
If a template is used by more than 25 pages, one of the following two errors can occur:
This is caused by a known exception in the invalidation message processing. The workaround is to turn "OFF" Web Cache in Oracle9iAS Portal.
For more information about troubleshooting Oracle9iAS Web Cache related issues in Oracle9iAS Portal, refer to chapter eight of the Oracle9iAS Portal Configuration Guide. This guide is available on OTN, see:
http://portalstudio.oracle.com.
Depending on the network configuration into which Oracle9iAS is installed, it may be necessary to disable IP checking in cookie validation.
This action is necessary if HTTP access to the Portal server is through a set of proxies that do not each assert the same IP address when calling Portal. In such cases, the login request that invokes the creation of the session cookie may be routed via one proxy, but subsequent requests are routed through different proxies. This raises a security violation during the IP checking step, and access to the page is therefore denied.
To change the state of IP checking in cookie validation, update data in both the Portal schema and the SSO schema as described in Table 4-1.
The following table has three columns, one heading row, and two body rows. Cells in column one indicate whether the desired action is enabling or disabling of IP checking. Cells in columns two and three describe the code entries that are necessary to update either the Portal schema or the SSO schema.
This information supersedes all other instructions in Oracle9iAS Portal documentation that describes how to disable IP checking during cookie validation, for example, Oracle9iAS Portal Release Notes, and Oracle9iAS Portal Configuration Guide Appendix B. These sections should now read as follows:
Oracle9iAS Portal Release Notes Section 5.5 -- IP Cookie Validation Failure
Some proxy server configurations may cause cookie validation to fail. This problem, probably only observed when Portal is being accessed from outside an intranet, such as from a mobile device, causes "Invalid Session" errors to appear in the client.
The workaround is to disable the IP checking stage of cookie validation by entering the SQL detailed earlier in the section "Disabling the IP Check of Cookie Validation".
Oracle9iAS Portal Release Notes Section 7.7 -- Need to Disable IP Checking if Accessing Mobile Pages
An error occurs when logging into Oracle9iAS Portal to access mobile-enabled pages on a wireless device. Turn off IP-checking during the authentication sequence.
To do this, enter the SQL detailed earlier in the section "Disabling the IP Check of Cookie Validation".
Oracle9iAS Portal Configuration Guide -Appendix B
To disable the IP check of cookie validation, enter the SQL detailed in the Portal Release Note section "Disabling the IP Check of Cookie Validation".
When creating a form based on a table or view (for a Portal DB provider), specify the table or view on which to base the form. Sometimes, if you click the LOV icon to display a list of available tables or views, the following message is displayed instead:
No Response from Application Web Server There was no response from the application web server for the page you requested. Please notify the site's webmaster and try your request again later.
The workaround is to enter the name (schema.objectname) directly into the "Table or View" field, for example, SCOTT.EMP.
It is not possible to create or copy items or pages using WebDAV clients. This issue causes a duplicate item to be created with the same name with nothing in it (0 bytes). It is possible to browse all of the standard pages and "file" type items that are created through the Oracle Portal UI. To fix this, apply a patch, which will be available on OracleMetalink at:
http://metalink.oracle.com
The following known issues are associated with Oracle9iAS Web Cache:
Oracle9iAS Web Cache may crash if the host name of a request URL is ambiguous, for example:
http://myhost:7777/ojspdemos/fragment.jsp.
To avoid potential problems, specify all possible variations of the site name with site aliases in the Site Definitions page (General Configuration > Site Definitions).
When Web Cache receives a request for a document, it looks at one of the following:
<esi:include> tag to determine the destination site
Web Cache then looks up the configured site settings and mappings (aliases) to determine if the site is supported. If the request does not include host information, then Web Cache sends the request to the default site.
The DAVOraWebCacheReadOnly parameter, if specified in the httpd.conf file, does not work with Web Cache version 9.0.2.0.0.
As a workaround, you can use the <LimitExcept> directive in the httpd.conf file, as described in the OraDAV module configuration chapter of the Oracle HTTP Server Administration Guide. Applying access restrictions to a location for all methods except GET, HEAD, and OPTIONS requests will essentially achieve the goal of using the DAVOraWebCacheReadOnly parameter. However, a caveat is that end users will always be restricted to GET, HEAD, and OPTIONS requests, even when Web Cache is not being used.
In the Oracle9iAS Web Cache Manager, you cannot change the wallet directory from "System Default Location". To change the wallet directory, you must make the following changes to ORACLE_HOME/webcache/webcache.xml:
CACHE element for your host.
OSWALLET element exists in webcache.xml (either as an element or an empty element), then change the text to the new wallet directory.
If no OSWALLET element exists, then you must create it between the RESOURCELIMITS and CALYPSONETINFO elements.
Example 4-1 shows a sample OSWALLET element.
<CACHE NAME="myhost.us.oracle.com-WebCache" ORACLEHOME="/u1/ora9ias" ...> ... <RESOURCELIMITS MAXINBOUNDCONNECTIONS="700" MAXCACHESIZE_MB="500"/> <OSWALLET>/u1/ora9ias/webcache/wallets/default</OSWALLET> <CALYPSONETINFO/> ... </CACHE>
For HTTPS requests, Internet Explorer 5.5 may send requests after Oracle9iAS Web Cache has already tried to close the connection. To resolve this issue, disable keep-alive by setting Keep-Alive to 0 on the Network Timeouts page (Cache-Specific Configuration > Network Timeouts) of the Oracle9iAS Web Cache Manager. This closes the connection between Oracle9iAS Web Cache and the client browser after the browser returns a response.
If you do not configure Oracle9iAS Web Cache during installation, and if you want to configure it at a later time using EMD, then Oracle9iAS Web Cache may fail to start. This might happen due to port conflicts with another Oracle9iAS Web Cache running on the same machine.
As a workaround, modify the Oracle9iAS Web Cache listen port number for SSL and non-SSL by logging into the Oracle9iAS Web Cache administration GUI and starting Oracle9iAS Web Cache. Because Oracle9iAS Web Cache was not configured during installation, some additional configuration steps are required to change the PORT directive in httpd.conf to the Oracle9iAS Web Cache port number.
Also, depending on the site configuration, change the Oracle9iAS Web Cache listen port value in ORACLE_HOME/webcache/webcache.xml and ORACLE_HOME/sysman/emd/targets.xml.
Avoid using the newgrp command on Linux Intel before installation to change the primary group. If you use newgrp before installation, the Web Cache Configuration Assistant might fail to start Web Cache after installation is complete. Using newgrp before installation creates a mismatch between the group identifier in the webcache.xml file (the primary group ID) and the group identifier of the installed files (the changed group ID).
If you must use newgrp, modify the webcache.xml file (line number 258) using the correct group ID after installation is complete, and run the following commands to start Web Cache:
$ cd $ORACLE_HOME/webcache/bin $ webcachectl start
You can ignore the webcache configuration assistant failure warning at the end of installation.
The following known issues are associated with Oracle9iAS Single Sign-On:
Logging out from Oracle9iAS Single Sign-On-enabled applications leaves non-Oracle9iAS Single Sign-On-enabled applications open, and logging out from non-Oracle9iAS Single Sign-On-enabled applications leaves Oracle9iAS Single Sign-On-enabled applications open.
Not all applications are integrated with Oracle9iAS Single Sign-On. If an application is not integrated with Oracle9iAS Single Sign-On, then its logout function is also not integrated. That means that when users click the logout button, they are logged out of that application only. They are not logged out of applications that are integrated with Oracle9iAS Single Sign-On. Users may mistakenly assume that they are logged out of all Oracle9i Application Server, which is not the case. Conversely, when users click the logout button from an Oracle9iAS Single Sign-On-enabled application, they are logged out of all Oracle9iAS Single Sign-On-enabled applications, but not out of applications that are not Oracle9iAS Single Sign-On-enabled.
The ssocfg.sh script is used to change the host, port, or protocol of the single sign-on server, as explained in Chapter 2 of the Oracle9iAS Single Sign-On Administrator's Guide. Script can be executed only if the environment variable LD_LIBRARY_PATH includes the value ORACLE_HOME/lib.
In Oracle9iAS, requests that are forwarded to the Single Sign-On Server for authentication can use only the HTTP method GET. If an application that is using OSSO utilizes methods other than GET, then the first page that is authenticated should be a GET. Also, the idle timeout feature should be disabled to insure that future requests that may potentially not be GET are not forwarded to the Single Sign-On Server.
The following known issues are associated with Oracle Internet Directory:
When you access the LDAP main page of EMD, the Oracle Internet Directory server may log extra information in a log file named oidldap00.log. This log file can be ignored.
In order to bring up the Oracle Internet Directory server on the default non-SSL port 389 and SSL port 636 at the end of the installation, the following conditions must hold true:
/etc/services file should not have any line containing these port numbers.
If either of these conditions is not met, then the Oracle Internet Directory server will be brought up on a different port which is logged in the file ORACLE_HOME/ldap/install/oidca.out.
In Oracle9iAS, you can install multiple Oracle9iAS components against a shared Oracle Internet Directory. In such a case, Oracle Internet Directory administrators might prefer to install Oracle9iAS Portal as an Oracle9iAS administrator so that they do not require Oracle Internet Directory super user privileges for such an installation. Currently, you can install most components and Oracle9iAS middle tiers if you have Oracle9iAS administrators privileges in Oracle Internet Directory. However, to perform an Oracle9iAS Portal installation, you will require Oracle Internet Directory super user privileges because an incorrect Access Control Policy in Oracle Internet Directory prevents you from installing as an Oracle9iAS administrator.
As a workaround, change the Access Control Policy on one of the nodes in Oracle Internet Directory. This should be done after Oracle Internet Directory has been installed and before you attempt to install Oracle9iAS Portal as an Oracle9iAS administrator. Also, the Oracle Internet Directory identity that is used to perform the Oracle9iAS Portal installation should be a member of the following group:
cn=OracleDASAdminGroup,cn=Groups,<DN of Default Subscriber Oracle Context> # Ldif script to change the ACL policy on Groups container. dn: cn=Groups, <Default Subscriber DN> changetype: modify replace: orclEntryLevelACI orclEntryLevelACI: access to entry by group="cn=oracledascreategroup, cn=groups,%s_OracleContextDN%" added_object_constraint=(objectclass=orclgroup) (browse, add) by group="cn=iASAdmins, cn=Groups,<DN of subscriber Oracle Context>" (browse, add) by * (none)
Perform the following steps to change the Access Control Policy:
<default Subscriber DN> in the ldif script above with the real DN of the default subscriber.
<DN of subscriber Oracle Context> to the DN of the subscriber Oracle Context. It is usually cn=OracleContext, <default Subscriber DN>.
temp.ldif.
ldapmodify command line tool to make the change in Oracle Internet Directory. For example:
ldapmodify -p <OID port> -h <OID host> -D <DN of OID Super User> -w <password of OID Super user> -v -f temp.ldif
The following are known issues associated with Oracle9iAS Containers for J2EE:
A customized version of the MERANT DataDirect Connect JDBC driver is shipped with Oracle9iAS to provide connectivity to non-Oracle databases. You should refer to standard MERANT documentation and release notes for technical information on the MERANT JDBC driver. However, it is important to note the differences between the standard MERANT JDBC driver and this customized version:
oracle" instead of "merant". When you connect, use the correct sub-protocol. For example:
jdbc:oracle:db2://server1:1433
com.oracle.ias (instead of com.merant.datadirect).
oias].
java.sql.SQLException: [oias][... JDBC Driver]
This driver is locked for use with embedded applications.
File README.TXT in the ORACLE_HOME/j2ee/home/jazn/demo/callerInfo directory has a reference to ORACLE_HOME/dcm/README, which does not exist. JAAS Readme is referenced to you from a core installation at:
http:<servername>/J2EE.htm
Perform the following steps to update the OC4J administrative password using the JAZN administration tool:
ORACLE_HOME environment variable is set and that you are using the correct java from the ORACLE_HOME/jdk/bin directory.
ORACLE_HOME/j2ee/home directory, use the following command to change the administrative password to the ias_admin password: (In the following example, the ias_admin password is "welcome1".)
java -Doracle.security.jazn.config=ORACLE_HOME/j2ee/home/jazn/install/jazn.xml -jar jazn.jar -setpasswd jazn.com admin welcome welcome1
java -Doracle.security.jazn.config=ORACLE_HOME/j2ee/home/jazn/install/jazn.xml -jar jazn.jar -checkpasswd jazn.com admin -pw welcome1
You should see "Successful verification of user/password pair".
The affected jazn-data.xml is located in the ORACLE_HOME/j2ee/home/config directory.
In order to take advantage of the java cache distributed functionality or to share a cached object between servlets, some minor modification to the deployment of an application may be necessary. Any user defined objects that will be shared between servlets or distributed between JVMs must be loaded by the system class loader. By default, objects loaded by a servlet are loaded by the context class loader. These objects are visible only to the servlets within the context that loaded them. The object definition is not available to other servlets or to the cache in another JVM. If the object is loaded by the system class loader, then the object definition will be available to other servlets and to the cache on other JVMs.
With Jserv, this was accomplished by including the cached object in the classpath definition available when the Jserv process was started.
With OC4J, the system classpath is derived from the manifest of the oc4j.jar file and from any associated .jars, including cache.jar. The classpath in the environment is ignored. To include a cached object in the classpath for OC4J, the .class file should be copied to ORACLE_HOME/javacache/sharedobjects/classes or added to the jar file ORACLE_HOME/javacache/cachedobjects/share.jar. The classes directory and the share.jar file have both been included in the manifest for cache.jar.
The following are known issues associated with Oracle9iAS Web Services:
A patch kit for UDDI v1.0 compliance of the Oracle9iAS UDDI registry can be downloaded from OracleMetalink at:
http://metalink.oracle.com
Follow the README.TXT in the top-level directory of the patch kit for installation and verification instructions.
When installed, this patch provides full UDDI v1.0 publishing support for the Oracle9iAS UDDI registry. Client applications can publish UDDI data, such as business information and WSDL for a Web service, to the Oracle9iAS UDDI registry programmatically via SOAP. For the convenience of client-side Java development environments, the Java APIs for UDDI are supplied with Oracle9iAS.
Oracle9iAS includes the full set of UDDI v1.0 publishing capabilities.
The README for Oracle9iAS SOAP stock quote sample is outdated. Keep in mind the following information when working with this sample:
To run stock quote sample behind the firewall:
dcmctl, define proxyHost and proxyPort in j2ee/home/config/oc4j.properties by adding the following lines:
proxyHost=<proxy host> proxyPort=<proxy port>
java -DproxyHost=<proxy host> -DproxyPort=<proxy port> -jar oc4j.jar
The following client programs for the Oracle SOAP Stored Procedure sample demo need to modified before compilation in order to run:
<ORACLE_HOME>/soap/samples/sp/company/ChangeSalary.java <ORACLE_HOME>/soap/samples/sp/company/GetAddress.java <ORACLE_HOME>/soap/samples/sp/company/GetEmp.java <ORACLE_HOME>/soap/samples/sp/company/GetEmpInfo.java <ORACLE_HOME>/soap/samples/sp/company/RemoveEmp.java
All of these programs use java.math.BigDecimal as one of the input parameters of the SOAP call. Any reference to java.math.BigDecimal.class should be changed to int.class.
You might see the following error message in the Oracle9iAS Containers for J2EE (OC4J) log file while running Java Messaging Service application:
# # HotSpot Virtual Machine Error, Internal Error # Please report this error at # http://java.sun.com/cgi-bin/bugreport.cgi # # Error ID: 43113F32554E54494D45110E43505002D3 # # Problematic Thread: prio=1 tid=0x6c40f320 nid=0x6f6d runnable #
This is a result of a known bug in JDK1.3.1_02 on Linux.
|
 Copyright © 2002 Oracle Corporation. All Rights Reserved. |
|