System problems can be divided into two basic categories:

• Cascading catastrophic failure

• Gradual breakdown in performance

Cascading catastrophic failure might be caused by:

• LDAP server is loaded and unresponsive

• Morning peak load starts

• Webgates send requests to the primary OAM Server

• Webgate requests time-out and Webgates retry to secondary OAM Server

Gradual breakdown in performance might occur over time when, for example:

• OAM is sized and rolled out for 10,000 users and 500 groups

• Over the course of a year, the number of users and groups increases significantly (to 50,000 users and 250 groups for example)

For information on the most commonly encountered issues, see the following topics:

• System Analysis and Problem Scenarios

• LDAP Server or Identity Store Issues

• OAM Server or Host Issues

• Agent-Side Configuration and Load Issues

• Runtime Database (Audit or Session Data) Issues

• Change Propagation or Activation Issues

• Policy Store Database Issues

This topic provides symptoms, probable cause, and steps to diagnose the following issues:

• Symptoms: Operational Slowness

• Symptoms: Total loss of service

Symptoms: Operational Slowness

• Poor user experience

• Agent time outs lead to retries

Cause

• Non-OAM load might be impacting OAM operations

• Capacity problems due to gradual increase in peak load

Symptoms: Total loss of service

Total loss of service

Cause

• Outage of all LDAP servers

• The load balancer is timing out old connections

Diagnosis

1. Shut down the LDAP server.

2. Restart your browser.

3. Try to access a protected site.

4. Review errors in the OAM Server log file, as described in Logging Component Event Messages (alternatively, in Monitoring Performance and Logs with Fusion Middleware Control).

5. Try to access Oracle Access Management Console.

6. Observe errors in WebLogic AdminServer log file.

7. Bring up the LDAP server again.

8. Retry access to a protected application.

9. Retry access to the Oracle Access Management Console.

10. Correct the issue based on the requirements in your environment.

This topic provides symptoms, probable cause, and steps to diagnose the following issues:

• Symptoms: Capacity Problems

• Symptoms: Interference with Other Services on the Host

Symptoms: Capacity Problems

• Poor user experience due to slow operations

• Agent time outs and retry can result in extra load

Cause

• CPU cycles

• Memory issues

Symptoms: Interference with Other Services on the Host

• Poor user experience due to slow operations

• Agent time outs and retry may result in extra load

Cause

• CPU cycle contention

• Memory contention

• File system full

Diagnosis: OAM Server

1. Shut down the OAM Server

2. Try to access a Webgate

3. Bring up the OAM Server

4. Use the Access Tester to test authentication and authorization as described in Validating Connectivity and Policies Using the Access Tester.

5. Use 'top' to figure out the CPU and Memory consumption of the OAM Server as you use the access tester

6. Get a thread dump of the OAM Server.

Diagnosis: AdminServer

1. Shut down the AdminServer.

2. Restart your browser and access a protected resource, which should work.

3. Use remote registration to register a new partner, as described in Registering and Managing OAM Agents (this should fail).

4. Startup OAM AdminServer.

This topic provides symptoms, probable cause, and steps to diagnose time issues between agents and servers.

Symptoms

Difference in Clock time Between Agent and Server

• High CPU usage at both agent and server

• User experiences a system hang

Cause

• Agent thinks the token issued by the server is invalid

• Agent keeps going back to the server to re-issue the token

Diagnosis

1. Access protected resource.

2. Confirm: Client access hangs.

3. Confirm: High CPU usage on agent and server.

Symptoms

• Audit and session operations are slow

• File system on the OAM Server is full with audit data that is not yet written to the database

• Loss of in-memory session data when one of the servers in the cluster fails

Cause

• Database is not tuned for write intensive operations

• Database is unavailable due to maintenance

• Space issues in the database

Diagnosis

1. Shut down the database used to store Audit and Session data.

2. Try to access a protected resource.

3. Review error and warning messages in the OAM Server log files, as described in Logging Component Event Messages (alternatively, in Monitoring Performance and Logs with Fusion Middleware Control).

This topic provides symptoms, probable cause, and steps to diagnose the following issues:

Symptoms

• Changes to policy do not take immediate effect

• Changes to system configuration do not take immediate effect

Cause

• Servers being too busy handling runtime requests (CPU contention)

Diagnosis

See "Policy Store Database Issues"

This topic provides symptoms, probable cause, and steps to diagnose policy database issues.

Symptoms

No policy changes are allowed; no impact on runtime

Cause

• Database is unavailable (down for maintenance)

• Space issues in the database

Diagnosis

1. Shut down the database containing OAM policies.

2. Try to access a protected resource and observe the runtime access is not impacted.

3. Try to access the Oracle Access Management Console to edit policies, and then observe errors in the AdminServer log file.

Problem

Challenge Redirect URL can be NULL; however, Challenge Method cannot be NULL.

If you open the Anonymous authentication scheme to edit, and click Apply without adding a value for Challenge method, the following errors might appear:

Messages for this page are listed below. 

* Challenge Method You must make at least one selection. 
 
* Challenge Redirect You must enter a value. 

Solution

You must include both a challenge method and a challenge redirect whenever you edit an anonymous authentication scheme.

The Access Manager X.509 Authentication Scheme relies on SSL to deliver the user's X.509 certificate to the OAM Server. The X.509 Authentication Scheme requires the X.509Plugin as the value of the Challenge Method (not the Authentication Module).

Problem

User has selected his certificate in the Browser but the Certificate is not available to the OAM Server.

Solution

The specific solution will depend on the reason for the SSL Handshake failure. For instance:

• For debugging SSL connections terminating on the Weblogic Server, please refer to http://docs.oracle.com/cd/E12840_01/wls/docs103/secmanage/ssl.html

• For debugging SSL connections terminating on the OHS server, see http://docs.oracle.com/cd/E12839_01/web.1111/e10144/under_mods.htm#i1007687.

Determine the reason for the SSL Handshake failure and the peer that is terminating the SSL Handshake. The solution will fall into the following categories:

• Configuration Issues

• Trust Issues

• Certificate Validation Issues

Problem

Single Sign Off might not work after accessing the resource with X.509 authentication. When the user is logged out with the logout URL and tries to access the resource in the same browser, authentication might not occur. Instead, the user should be asked for authentication using the certificate pop up.

This can occur with any Agent type.

Solution

After executing the logout URL, click on Clear SSL State from the browser as follows, and the access the X.509-protected resource:

From the browser window, open the Tools menu, click Internet Options, choose Content, and then Clear SSL state.

Problem

Client certificate authentication works fine using the standard X509 Authentication Module after importing the root and sub CA certificates into the WebLogic Server and .oamkeystore keystores.

However, a certificate validation error can occur when using a Custom X509Plugin Authentication Module and root and sub CA certificates into the WebLogic Server and .oamkeystore keystores.

Solution

With the Custom X509Plugin Authentication Module the root and sub CA certificates must be added to the DOMAIN_HOME/config/fmwconfig/amtruststore because the X509CredentialExtractor plug-in loads certificates from this location.

An error is logged in the oam-server diagnostic log file whenever you create or edit an IPv4 range or temporal condition:

.... refreshPolicy specified but no response collector supplied 

Cause

This is a message that is erroneously being logged at the ERROR level.

Solution

The correct level of the message is INFO.

If too many results are returned, you are informed as follows:

Description of the illustration id_ldapfilter_results.gif

Solution

1. Click OK.

2. Click Test Filter to initiate a new test.

3. In the Edit Search Filter dialog, make your changes.

4. Check the Test Results.

Some characters might not be usable within header response names or values, depending on whether the client receiving these responses is a Webgate, and if so which Web server is protected. Certain characters might be subject to automatic conversion to other characters in a server-specific way.

Oracle recommends that you refer to your Web server documentation for more details.

If you attempt to create and apply configuration details for an OAM Server before configuring the OAM Server in the WebLogic Server domain, a message informs you of the following:

Configuration does not exist for path
/DeployedComponent/Server/oamServer/Instance/test

For more information, please see the server's error log for
an entry beginning with: Server Exception during PPR, #6.

To resolve this issue, you must configure the OAM Server in the WebLogic Server domain before you register the configuration with Access Manager.

If the number of requests to the OAM Server unexpectedly increases beyond what the server can handle, it could crash.

To limit the number of requests to the OAM Server:

1. In the WebLogic Console, use the Message Driven Bean pool to restrict the number of NAP requests to the OAM Server.

   MDBeans pull NAP requests from the Server queue and deliver NAP requests to the Server for processing. Limiting the number of MDBean instances helps control the number of requests that are processed at a given time.

2. In the WebLogic Console, configure the number of WebLogic worker threads that can be used (to restrict the number of requests to the OAM Server).

   MDBeans pull NAP requests from the server queue and deliver NAP requests to the Server for processing. Limiting the number of MDBean instances helps control the number of requests that are processed at a given time.

3. In the WebLogic Console, configure the number of WebLogic worker threads that can be used (to restrict the number of requests to the OAM Server).

   See the topic on Thread Management in the guide to Oracle Fusion Middleware Performance and Tuning for Oracle WebLogic Server.

4. In the WebLogic Console, specify a maximum incoming request size, complete message timeout, and set the number of file descriptors, to optimize performance as described in following topics in the Oracle Fusion Middleware Performance and Tuning for Oracle WebLogic Server:

   • Tuning Message Size

   • Tuning Complete Message Timeout

   • Tuning Number of File Descriptors

Consider the scenario where Webgate sends an authentication request to the OAM Server. After successful credential collection and validation, the OAM Server creates the session and the relevant cookies (OAM_ID, ObSSOCookie). However, due to network latency, the response times out by the time the OAM Server sends it to the Webgate which triggers Webgate to re-send the authentication request to the Server. The OAM Server recognizes the session, then recreates the ObSSOCookie, and sends the response to the agent. If the network latency persists, the cycle continues in an infinite loop between the Server and the Webgate. The user is neither asked to login again nor presented with an error message.

ModSecurity is a Web application firewall (WAF) that can be deployed as part of the existing Apache-based Web server infrastructure. This module can be plugged into the OHS Server that front-ends the OAM Server. In this way, Mod_security module protects the OAM Server from denial of service attacks.

A flexible rule engine is at the heart of ModSecurity. It implements the ModSecurity Rule Language, a specialized programming language designed to work with HTTP transaction data. A new configuration directive uses the httpd-guardian script to monitor for Denial of Service (DoS) attacks. By default httpd-guardian defends against clients that send more than 120 requests in a minute, or more than 360 requests in five minutes.

To protect from a flood of HTTP Requests

1. Add the mod_security module to the OHS Server that front-ends the OAM Server.

2. In the OHS Server configuration, set the configuration directive to use the httpd-guardian script to monitor for Denial of Service (DoS) attacks.

   Syntax:

   SecGuardianLog |/path/to/httpd-guardian
   

   Example:

   SecGuardianLog |/usr/local/apache/bin/httpd-guardian
   

Problem

OAM Server does not start up.

Solution

1. Locate and review the OAM Server log file on the computer hosting the OAM Server.

    DOMAIN_HOME/servers/SERVER-NAME/logs/SERVER-NAME-diagnostics.log
   

2. Enable logging for this computer, as described in Logging Component Event Messages:

    DOMAIN_HOME/config/fmwconfig/servers/SERVER-NAME/logging.xml
   

3. Restart the OAM Server, observe the behavior, check the log file again if needed.

Problem

Monitoring the OAM Server reveals a significant spike in latency during authentication.

Solution

1. Locate and review the OAM Server log file on the computer hosting the OAM Server.

    DOMAIN_HOME/servers/SERVER-NAME/logs/SERVER-NAME-diagnostics.log
   

2. Enable logging for this computer, as described in Logging Component Event Messages:

    DOMAIN_HOME/config/fmwconfig/servers/SERVER-NAME/logging.xml
   

3. Restart the OAM Server, observe the behavior, check the log file again if needed.

Problem

Debugging for all expression parsing and evaluation produced a significant performance drag within ~20 hours due to memory growth; running out of memory in ~50 hours.

Configuration: 2GB heap; 3 minute session timeout; jdbc connections tuned min=32 max=200; jdbc connection idle timeout disabled; jbo pool size min = 10 & max=150

Solution

To generate heap-dumps for comparison, you use the following command-line tools jmap for Sun jvm or jrcmd for jrockit jvm located under JAVA_HOME/bin.

For jrockit jvm

jrcmd pid <command> 
/jrockit_160_14_R27.6.5-32/bin/jrcmd 16775 heap_diagnostics
/jrockit_160_14_R27.6.5-32/bin/jrcmd 16775 print_threads
/jrockit_160_14_R27.6.5-32/bin/jrcmd 16775 jrarecording ....

For Sun jvm

jmap -histo <pid> 
jmap -dump:live,format=b,file=heap.bin <pid>

If form authentication or pass-through functionality is not working, the problem might be that either "UseWebGateExtForPassthrough" parameter is not set to true in the Webgate profile or that webgate.dll is not configured as Wild Card Application Mapping in IIS. In such cases, Webgate does not perform authentication or authorization for HTTP "POST" requests for the resources protected by form-based authentication.

Solution

Confirm that the UseWebGateExtForPassthrough parameter is configured in the Webgate profile with a value of true and that webgate.dll is configured as Wild Card Application Mapping.

Following are some general guidelines to follow when installing Access Manager Webgates with IIS Web servers.

Account Privileges: The account that performs Access Manager installation must have administration privileges. The user account that is used to run OAM services must have the "Log on as a service" right, which can be set by selecting Administrative Tools, Local Policy, Local Policies, User Rights Assignments, Log on as a service.

IIS 6 Web Servers: You must run the WWW service in IIS 5.0 isolation mode. This is required by the ISAPI postgate filter. During Access Manager installation, this is usually set automatically. If it is not, you must set it manually for the Default Web site.

Webgate for IIS 7 Web Server: To use Form-based authentication without enabling pass through functionality (for example, "access/oblix/apps/webgate/bin/webgate.dll" is an action in the Form-based authentication scheme), ensure that the entry "<add segment="bin"/>" is not present in the applicationHost.config file. If the entry is present, you must remove it. Use the following steps to check this entry:

• Go to Windows\System32\inetsrv\config and open the file applicationHost.config.

• Search for the <hiddenSegments> module and remove the entry <add segment="bin"/> if it is present.

Webgate: When installing IIS Webgates, setting various permissions for the /access directory is required for IIS Webgates only when you are installing on a file system that supports NTFS. For example, suppose you install the ISAPI Webgate in Simple or Cert mode on a Windows 2000 computer running the FAT32 file system. The last installation panel provides instructions for manually setting various permissions that cannot be set on the FAT32 file system. In this case, these instructions may be ignored.

On IIS 6 Web servers only, you must run the WWW service in IIS 5.0 isolation mode, which is a requirement of the ISAPI postgate filter. This scenario will work if you have 32-bit Access Manager binaries running on a 32-bit Windows operating system. However, there is an issue if you attempt to run a 32-bit postgate.dll on a 64-bit Windows machine with IIS running in 32-bit mode.

Problem

When running IIS in IIS5.0 isolation mode, you see the following message:

"ISAPI Filter 'C:\webgate\access\oblix\apps\webgate\bin\webgate.dll' could not be loaded due to a configuration problem.

Cause

The current configuration only supports loading images built for an AMD 64-bit processor architecture. The data field contains the error number.

Solution

To learn more about this issue, including how to troubleshoot this kind of processor architecture mismatch error, see the following Web site:

http://go.microsoft.com/fwlink/?LinkId=29349

For more information, see Help and Support Center at:

http://go.microsoft.com/fwlink/events.asp

Problem

IIS5 never existed as 64-bit. However, IIS v6's IIS5 Compatibility Mode on 64-bit Windows computers only runs as 64-bit.

Cause

It is architecturally impossible run IIS5 Isolation Mode 32- bit on 64-bit Windows, as described in documentation available through the following URLs:

http://www.microsoft.com/communities/newsgroups/en-us/default.aspx?dg=microsoft.pu
blic.inetserver.iis&tid=5dd07102-8896-40cc-86cb-809060fa9426&cat=en_US_
02ceb021-bb43-476d-8f8f-6c00a363ccf5&lang=en&cr=US&p=1

http://blogs.msdn.com/david.wang/archive/2005/12/14/HOWTO-Diagnose-one-cause-of-W3
SVC-failing-to-start-with-Win32-Error-193-on-64bit-Windows.aspx

A "The page cannot be displayed" error that appears after configuring Webgate for pass-through functionality, indicates a configuration issue.

Solution: Confirm that the UseWebGateExtForPassthrough parameter is configured in the Webgate profile with a value of true and that webgate.dll is configured as Wild Card Application Mapping.

When Access Manager is running with Microsoft's IIS Web server, you must manually uninstall and reinstall the following ISAPI filters when reinstalling Access Manager.

• tranfilter.dll

• oblixlock.dll (if you installed Webgate)

• webgate.dll (if you installed Webgate)

To remove and reinstall IIS DLLs

1. Uninstall Access Manager.

2. Manually uninstall the preceding DLLs.

3. Reinstall Access Manager.Active Directory.

4. Manually reinstall the DLLs.

The automatically generated Description for some components are not translated. This is expected and enables Administrators to change the Description to whatever they require. Following such a change, translation by Oracle is not possible.

The Oracle Access Management Console displays policies and resources oddly when the input configuration file for remote registration is not in UTF-8 format or when the OAM Server is not started in UTF-8 locale (en_US.utf8, for instance).

Be sure to use UTF-8 encoding if creating a configuration file for the remote registration tool, oamreg, to generate authentication policies and protected resources. Also, be sure to start OAM Server in UTF-8 locale machines. Otherwise, the Oracle Access Management Console might display policies and resources oddly following successful inband registration.

Configure Access Manager to use Kerberos Authentication Scheme with WNA challenge method, and create a non-ASCII user in Microsoft Active Directory.

Problem

An exception occurs when trying to get user details to populate the subject with the user DN and GUID attributes. Authentication fails and an error is recorded in the OAM Server log when a non-ASCII user in Active Directory attempts to access an Access Manager-protected resource:

... Failure getting users by attribute : cn, value ....

Cause

The username in the attribute is passed without modification as a java string.

Solution

Non-ASCII users can access the resource protected by Kerberos WNA scheme now by applying this JVM system property (for the built-in WebLogic SPNEGO support):

-Dsun.security.krb5.msinterop.kstring=true

Register a Webgate with Access Manager using a non-ASCII name. In the Access Tester, enter the valid IP Address, Port, and Agent ID (non-ASCII name), then click Connect.

Connection testing fails.

When the browser locale is not supported, the Oracle Access Management Console Login page shows as server locale. It should fall back to English. This is the expected behavior:

• If the client Locale is not supported, Oracle Access Management falls back to the server locale.

• If the server locale is not supported, Oracle Access Management falls back to English.

When users select an unsupported language and come to the Access Manager SSO page, it shows as server locale (German, for example). However, after logging in, all the pages are displayed as English.

To fall back to English

Disable the Access Manager SSO page and the original Access Manager login page also falls back to English.

Solution

Ensure that the agent name is unique (does not already exist) and that the AdminServer is running.

Solution

Protected and public resources must be described as relative URLs of the format '/index.html'. If the resource does not begin with a '/', no ObAccessClient.xml file will be generated.Verify the protected and public resource URLs and ensure all begin with a "/". For more information, see "Resource URL, Prefixes, and Patterns".

Partner registration can fail if you do not supply a unique agent name, which is also used to create an Application Domain. The agent name and Application Domain name must be the same and must be unique. Using the oamreg validate command can fail when the agent name does not match the Application Domain name.

Solution

Ensure that the agent name and Application Domain name are the same.

Session impersonation is not enabled by default. You can update the value in oam-config.xml, then update the version of oam-config.xml to automatically propagate the ImpersonationConfig status to all managed servers without a restart.

To enable Session Impersonation

1. Back up DOMAIN_HOME/config/fmwconfig/oam-config.xml.

2. Set ImpersonationConfig to true:

   <Setting Name="ImpersonationConfig" Type="htf:map">
        <Setting Name="EnableImpersonation" Type="xsd:boolean">false</Setting> 
   </Setting>      
   

3. Configuration Version: Increment the Version xsd:integer as shown in the next to last line of this example (existing value (25, here) + 1):

   Example:

   <Setting Name="Version" Type="xsd:integer">
     <Setting xmlns="http://www.w3.org/2001/XMLSchema"
       Name="NGAMConfiguration" Type="htf:map:> 
     <Setting Name="ProductRelease" Type="xsd:string">11.1.1.3</Setting>
       <Setting Name="Version" Type="xsd:integer">25</Setting>
   </Setting>      
   

4. Save oam-config.xml.

Expected Behavior: Oracle Identity Federation 11.1.1 session is not cleared

When Oracle Access Manager 11.1.1 is integrated with Oracle Identity Federation 11.1.1, and you clear the session using the console, only the Oracle Access Manager session is cleared. The Oracle Identity Federation session is not cleared.

Problem

AdminServer start up fails with following message:

"java.net.SocketException: 
No buffer space available".

Configuration for the number of AIX file descriptors set for the operating system is substantially high (ulimit file descriptor) resulting in a buffer overflow that causes remote registration failure with the following message:

The ulimit value is application dependent and applies exclusively to application program data and the application stack. The default number of open files setting (2000) is typically sufficient for most applications. If the value is too low, errors might occur when opening files or establishing connections. Because this value limits the number of file descriptors that a server process might open, a value that is too low prevents optimum performance. For the AIX operating system, the default setting is 2000.

Solution

Increasing the ulimit file descriptor limits might improve performance. Increasing some of the other limits might be needed depending on your application.

1. Log in as root.

2. Perform the following steps to change the open file limit to 10,000 files:

   1. Open the command window.

   2. Locate and edit /etc/security/limits file to add the following lines to the user account on which the AdminServer process runs:

      nofiles =  10000   
      nofiles_hard = 10000
      

   3. Save the file and restart AIX.

3. In a command window, decrease the TCP_TIMEWAIT interval with the following command to set the state to 15 seconds (which allows TCP to release closed connections faster and increases the number of available resources for open connections).

   /usr/sbin/no –o tcp_timewait =1
   

4. Tune the following parameters to 256k, as shown:

   no -a |grep space   
         tcp_recvspace = 262144   
         tcp_sendspace = 262144   
         udp_recvspace = 262144  
         udp_sendspace = 262144
   

5. Tune the following parameters as indicated here:

   no -o rfc1323=1  
   no -o sb_max=4194304

Problem

While creating an Authentication or Authorization Policy, if you add a resource that is already used in another Authentication or Authorization Policy, a validation error appears when you click Apply. This is expected.

If you click OK in the error window and then attempt to add a valid resource that is not used within another Authentication or Authorization Policy, the resource is not added and the Authentication or Authorization Policy is not created.

Solution

1. Click Apply and close the Authentication or Authorization Policy page.

2. From the navigation tree, click the named policy again, click the Edit to open the page, and add the new resource.

Problem

A validation error appears if you enter an optional description longer than 200 characters.

Solution

Keep optional descriptions to 200 characters in length and less than 10 lines.

Symptom

You are running an Apache Web server, and an OAM Server fails, displaying the following message:

libthread panic: cannot create new lwp
(PID: 9035 LWP 2). stackrace:
ff3424cc
0

This symptom may be caused by the Apache Web server launching more instances of itself. This can happen when the server determines that more instances are needed to service the number of connections between one or more Webgates and the OAM Server.

The additional instances create even more connections, which exceed the number of connections by the OAM Server.

Solution

Reduce the number of MinSpareServers, MaxSpareServers, StartServers, and MaxClients parameters.

Go to the OAM Server's configuration directory and open the http.d configuration file.

Recommended parameter settings:

• MinSpareServers 1

• MaxSpareServers 5

• StartServers 3

• MaxClients 5

When running Apache v2 on HP-UX, do not use nobody for User or Group, because shared memory may not work. Instead, use your login name as User Name with a your group as Group Name On HP-UX (on Solaris, "www" is equivalent to "nobody").

When running Apache v2 on HPUX 11.11, ensure that the AcceptMutex directive in the Apache httpd.conf file is set to "fcntl". If the directive is not present, add it to the httpd.conf file (AcceptMutex fcntl). For more information, see:

http://issues.apache.org/bugzilla/show_bug.cgi?id=22484

Problem

After installing a Webgate on vendor-bundled Apache, the Web server may give the following error upon startup:

Error: Cannot load libgcc_s.so.1 library - Permission denied. 

Solution

Change the Security-Enhanced Linux (SELinux) policy rules for Access Manager Webgates.

Errors might be reported in WebServer logs/console when starting a Web server on Linux distributions, which have stricter SELinux policies in place, after installing an Access Manager Web component. You can avoid these errors by running appropriate chcon commands for the installed Web component before restarting the Web server.

The following item is required only if you compile Apache v2 for Webgate on UNIX with the mpm_worker_module. In this case, you need to modify the thread.c file from the Apache source for the UNIX environment. Making this change ensures that the default pthread stacksize for Webgate produces optimal performance during multi-threaded server implementation. If this change is not made, the default pthread stack size would not be sufficient for Webgate and could result in a crash.

Apache 2.0 does not support the ThreadStackSize option. Therefore:

• With UNIX-based Apache v2.1 and later you must use the ThreadStackSize directive to set the size of the stack (for autodata) of threads that handle client connections and call modules to help process those connections.

• With UNIX-based Apache 2, it is best to use the compilable source while adding the mpm_worker_module and changing the thread.c file to avoid a stack overflow.

The following procedure shows how to modify the Apache v2.0 thread.c file to provide the default pthread stacksize needed by Webgate for optimal performance during multi-threaded server implementation. For details about the Apache v2.1+ ThreadStackSize directive, see http://httpd.apache.org/docs/2.2/mod/mpm_common.html#threadstacksize.

To modify the Apache v2.0 thread.c file for Webgate in a UNIX environment

1. Locate the thread.c file. For example:

   APACHE 2.0.52 source/srclib/apr/threadproc/unix/thread.c
   

2. Locate the function named apr_threadattr_create(apr_threadattr_t **new,apr_pool_t *pool) in the following code segment:

   **new,apr_pool_t *pool) in the following code segment:
   1-----> apr_status_t stat; 
   2 
   3-----> (*new) = (apr_threadattr_t *)apr_pcalloc(pool, sizeof(apr_threadattr_t)); 
   4-----> (*new)->attr = (pthread_attr_t *)apr_pcalloc(pool, sizeof(pthread_attr_t)); 
   5 
   6-----> if ((*new) == NULL || (*new)->attr == NULL) { 
   7----->              return APR_ENOMEM; 
   8-----> } 
   9 
   10----->(*new)->pool = pool; 
   11----->stat = pthread_attr_init((*new)->attr); 
   12 
   13-----> if (stat == 0) { 
   14----->            return APR_SUCCESS; 
   15-----> } 
   16----->#ifdef PTHREAD_SETS_ERRNO 
   17----->stat = errno; 
   18----->#endif 
   19 
   20----->return stat; 
   21 
   

3. Add the following code before line 13 shown earlier.

   int stacksize = 1 << 20; 
   pthread_attr_setstacksize(&(*new)->attr, stacksize);
   

4. Run configure, make, and make install to set up the Apache Web server with the mpm_worker_module.

Failure Authentication Event: For Domino Web servers, the redirection of a URL through Access Manager may not work if the authentication type is set as Basic Over LDAP and the URL to be redirected is mentioned as one of the following:

• Either a relative path present on the same Web server
• Or the Full path URL on the same Web server containing a computer name defined in the host identifier string combinations.

To overcome a failure authentication event, you must set the redirected URL with a computer name that is not defined under the host identifier group. For example, the IP address of the computer.

This problem does not occur with a form-based authentication type.

Header Variables: It may not be possible to pass header variables other than REMOTE_USER to Webgates installed on Lotus Notes Domino Web servers when using Client Certificate authentication scheme.

For example, header variables cannot be set on the one request where Client Certificate authentication occurs. However, all other requests do allow header variables to be set.

Symptom

If you installed Access Manager on UNIX under a different user ID than you used to create your Web server instance, Access Manager can become unstable. Users may experience behavior such as:

• Random bug report pages

• Failure to write to log file errors

• Loss of access to Web pages

Solution

Change file permissions using the chown command. Change the Access Manager directory to the same user ID that you used to create your Web server instance.

Webgate uses ISAPI extension for displaying user deny error message and for displaying the diagnostic page. However, ISA 2006 does not support extensions. Therefore:

• If the user is denied access by Webgate, the user gets Page Cannot be displayed error message instead of Access Manager denied access error message.

• The following diagnostic URL does not work for ISA: http(s)://hostname:port/access/oblix/apps/webgate/bin/webgate.dll?progid=1 for webgate.

Problem

After installing a Webgate instance on an Oracle HTTP Server, the server does not start up.

Cause

This occurs because Access Manager uses an older Linux threading model.

Solution

When using LinuxThreads mode, comment out the Perl module in the httpd.conf file, update the LD_ASSUME_KERNEL environment variable, and restart, as described in the following procedure.

To resolve the failure to start Oracle HTTP Server in LinuxThreads mode

1. Comment out the perl module in the httpd.conf file in the following location:

   Oracle HTTP Server 11g: $ORACLE_INSTANCE/config/OHS/ohs_name/httpd.conf

   Oracle HTTP Server v2: OH$/ohs/conf/httpd.conf

   Oracle HTTP Server v1.3: OH$/Apache/Apache/conf/httpd.conf

2. To update the LD_ASSUME_KERNEL value, open the following file in a text editor:

   OH$/opmn/conf/opm.xml
   

3. Find the following line:

   <process-type id="HTTP_Server" module-id="OHS">

4. Add the following information under the line you found in the previous step:

   <environment>
   <variable id="LD_ASSUME_KERNEL" value="2.4.19" />
   </environment>
   

5. Save this file.

6. Run the following commands to implement your changes:

   opmnctl stopall
   opmnctl startall

This situation might arise whether you are using Access Manager with LinuxThreads or NPTL.

Symptom

Webgate fails to initialize when installed on an Oracle HTTP Server running Red Hat Enterprise Server version 4.0 with a kernel version lower than 2.6.9-34.EL. Version 2.6.9-34.EL is supplied with the Red Hat version 4, update 3.

Solution

To prevent this problem, you must upgrade to Red Hat version 4, update 3 or higher.

Problem

With Oracle Application Server 10.1.x, OC4J, when the httpd.conf file is modified automatically during Webgate installation, it can be corrupted.

Solution

Before installing Webgate, run the following command to prevent the httpd.conf file from being overwritten.

$ORACLE_HOME/dcm/bin/dcmctl updateConfig -ct ohs 

On IIS 6 Web servers only, you must run the WWW service in IIS 5.0 isolation mode, which is a requirement of the ISAPI postgate filter. This scenario will work if you have 32-bit Access Manager binaries running on a 32-bit Windows operating system. However, there is an issue if you attempt to run a 32-bit postgate.dll on a 64-bit Windows machine with IIS running in 32-bit mode.

Problem

When running IIS in IIS5.0 isolation mode, you see the following message:

"ISAPI Filter 'C:\webgate\access\oblix\apps\webgate\bin\webgate.dll' could not be loaded due to a configuration problem.

Cause

The current configuration only supports loading images built for an AMD 64-bit processor architecture. The data field contains the error number.

Solution

To learn more about this issue, including how to troubleshoot this kind of processor architecture mismatch error, see the following Web site:

http://go.microsoft.com/fwlink/?LinkId=29349

For more information, see Help and Support Center at:

http://go.microsoft.com/fwlink/events.asp

Problem

IIS5 never existed as 64-bit. However, IIS v6's IIS5 Compatibility Mode on 64-bit Windows computers only runs as 64-bit.

Cause

It is architecturally impossible run IIS5 Isolation Mode 32- bit on 64-bit Windows, as described in documentation available through the following URLs:

http://www.microsoft.com/communities/newsgroups/en-us/default.aspx?dg=microsoft.pu
blic.inetserver.iis&tid=5dd07102-8896-40cc-86cb-809060fa9426&cat=en_US_
02ceb021-bb43-476d-8f8f-6c00a363ccf5&lang=en&cr=US&p=1

http://blogs.msdn.com/david.wang/archive/2005/12/14/HOWTO-Diagnose-one-cause-of-W3
SVC-failing-to-start-with-Win32-Error-193-on-64bit-Windows.aspx

Symptom

When attempting to start the Sun Web server, you get an error like the following:

Unable to start, PCLOSE

Solution

Solution: A number of problems can cause this error:

• A syntax error in your obj.conf file

• Leading spaces in your obj.conf file

• Installing Access Manager as a different user ID than what you used to create your Web server instance

• A carriage return at the end of the obj.conf file

When Access Manager is running with Microsoft's IIS Web server, you must manually uninstall and reinstall the following ISAPI filters when reinstalling Access Manager.

• tranfilter.dll

• oblixlock.dll (if you installed Webgate)

• webgate.dll (if you installed Webgate)

To remove and reinstall IIS DLLs

1. Uninstall Access Manager.

2. Manually uninstall the preceding DLLs.

3. Reinstall Access Manager.Active Directory.

4. Manually reinstall the DLLs.

Description

This command enables Multi-Data Center mode. It takes a value equal to the full path to, and name of, the MDC.properties file.

Syntax

enableMultiDataCentreMode(propfile="../MDC_properties/oamMDCProperty.properties")

Sample oamMDCProperty.properties File

SessionMustBeAnchoredToDataCenterServicingUser=false
SessionDataRetrievalOnDemand=true
Reauthenticate=true
SessionDataRetrievalOnDemandMax_retry_attempts=3
SessionDataRetrievalOnDemandMax_conn_wait_time=80
SessionContinuationOnSyncFailure=true

#MDCGitoCookieDomain=.example.com <This setting should be provided only if there is a common cookie subdomain across the WGs and DCs>

Example

The following command enables this data center.

enableMultiDataCentreMode(propfile="../MDC_properties/oamMDCProperty.properties")

Description

This command disables Multi-Data Center mode.

Syntax

disableMultiDataCentreMode()

There are no arguments for this command.

Example

The following command disables Multi-Data Center mode.

disableMultiDataCentreMode()

Description

This command adds a partner to the Data Center. It takes a value equal to the full path to, and name of, the partnerInfo.properties file.

Syntax

addPartnerForMultiDataCentre(propfile="../MDC_properties/partnerInfo.properties")

Table G-2 documents the properties that comprise partnerInfo.properties. See Multi-Data Center Security Modes for properties file samples.

Example

The following command defines this data center as a Master.

addPartnerForMultiDataCentre(propfile="../MDC_properties/partnerInfo.properties")

Description

This command removes a registered remote partner from a configured Data Center. It takes a value equal to a valid remoteDataCentreClusterId.

Syntax

removePartnerForMultiDataCentre=("<cluster_ID>")

Example

The following command defines the partner to be removed.

removePartnerForMultiDataCentre("99bf9-adc2120609")

Description

In an MDC deployment one Data Center is designated as the Master and the others as a Clone. Essentially all MDC wide global configurations and policy updates should be applied to the Master and propagated to the Clones. This command sets the type of the data center. Values are Master or Clone.

Syntax

setMultiDataCenterType(DataCenterType="<Master|Clone>")

Example

The following command defines this data center as a Master.

setMultiDataCenterType(DataCenterType="Master")

Description

A Clone Data Center can be write protected by setting WriteEnabledFlag to false. In this case, the Clone Data Center will not allow updates through the Oracle Access Management Console or WLST commands. Data synchronization will still continue to update as the command is used to write protect the Clone Data Center against accidental updates after the initial set up is complete.

Syntax

setMultiDataCenterWrite(WriteEnabledFlag="<true|false>")

Example

The following example protects the Clone Data Center from accidental overwrites.

setMultiDataCenterWrite(WriteEnabledFlag = "false")

Description

This command sets the Multi-Data Center cluster name. Value is a string.

Syntax

setMultiDataCentreClusterName(clusterName="<string_value>")

Example

The following command enables this data center.

setMultiDataCentreClusterName(clusterName="MyCluster")

Description

This command validates that the required entries in the Multi-Data Center configuration are present in oam-config.xml. For the MDC solution, a new Access Manager event named mdc_session_update is required to create or update MDC sessions during authorization. The Access Manager event model requires a set of configurations to be present in the oam-config.xml configuration file. The required configurations cannot be added statically so validateMDCConfig validates the required entries for mdc_session_update and seeds any configurations not already present.

Syntax

validateMDCConfig()

There are no arguments for this command.

Example

The following command validates the MDC configuration.

validateMDCConfig()

Description

This command will create a ZIP file of the Master Data Center UDM metadata.

Syntax

exportAccessStore(toFile="<name and location of ZIP", namePath="/")

Example

exportAccessStore(toFile="/master/location/dc1metadata.zip", namePath="/")

Description

This command will import a ZIP file of the Master Data Center UDM metadata to the Clone Data Center.

Syntax

importAccessStore(fromFile="<name and location of ZIP", namePath="/")

Example

importAccessStore(fromFile="/master/location/dc1metadata.zip", namePath="/")