Skip navigation.

Release Notes

  Previous Next vertical dots separating previous/next from contents/index/pdf Contents View as PDF   Get Adobe Reader

WebLogic Server 8.1 Known Issues

The following sections describe known issues, levels of support, notes, and general problems in WebLogic Server 8.1 and its service packs:

 


Known Issues in WebLogic Server 8.1 SP6

Core Known Issues

Change Request Number

Description

CR263825

You may experience performance issues due to changes in JRockit SocketMuxer support for JRockit 1.4.2_10 and all later 1.4.2 releases.

Workaround:

Contact your BEA support representative for the CR263825_810sp6 patch.

JDBC Known Issues

Change Request Number

Description

CR280439

There is a performance degradation in the WebLogic Type 4 JDBC Driver for Oracle getAsciiStream() method call caused by fixes and enhancements to character set support in the driver. In general, the updated driver outperforms the previous version of the driver.

Workaround or Solution:

Use the getCharacterStream() method instead of getAsciiStream().

JTA Known Issues

Change Request Number

Description

CR259037

When a transaction times out at logging time, you could experience a deadlock situation. The thread that is trying to write the transaction to the transaction log (TLOG) times out, a different thread trying to execute the rollback to clean the same transaction from the TLOG and deadlock each other.

Workaround or Solution:

Decrease the value of the CheckpointIntervalSeconds attribute and move TLOGs to a directory where they will not run out of memory.

Servlets & JSPs Known Issues

Change Request Number

Description

CR262289

When calling the close() method on the InputStream obtained from the HttpsURLConnection class, the underlying SSL socket might not be closed even when the disconnect() method is called. This can result in a socket leak.

Workaround or Solution:

To avoid such a socket leak and ensure all underlying resources are closed, you are advised to use the HttpsURLConnection#disconnect() method instead of the InputStream#close() method.


 

Console Help Know Issues

Change Request Number

Description

CR227849

Unable to start WebLogic Server after changing the password from the Administration Console.

Workaround or Solution:

After you change the user password from the Administration Console, make sure that you also change the password in the <domain directory>/boot.properties file. In this file, the password is stored in the encrypted format. However, when you edit the file, you can enter the password in clear text. After restarting the server, the password is automatically encrypted and stored in the file.

Note: Updating the password in the boot.properties file is applicable to both Administration Server and Managed Servers.

To remotely start Managed Servers from the Administration Console, after changing the password, in addition to changing the password in the boot.properties file, make sure that you change the password from the Remote Start tab in the Administration Console. For information about changing the password from the console, see "Changing the Password of a User" in Security in Administration Console Online Help.


 

 


Known Issues in WebLogic Server 8.1 SP5

Core WebLogic Server Known Issues

Change
Request
Number

Description

CR185820 CR185821

If a client has installed a reverse-proxy server with IP mapping, server IP addresses and domain names may be unavailable.

CR217067

The Low Memory Granularity Level attribute is disabled. WebLogic Server logs a warning message when the average free memory value is smaller than the Low Memory GC Threshold value.

CR236083

Clustered HttpSession replication may trigger a "Received a stale replication request" error when all of the following conditions exist:

    1. The cluster is configured with PersistentStoreType set to replicated (and/or replicated if clustered).

    2. The cluster is replicating HTTP sessions for a heavy client load, with several hundred concurrent users.

    3. Some server threads handling client traffic are delayed while waiting for responses from a database.

CR272727

The ulimit -n command to limit the number of open file descriptors is not supported in ksh shell on HP-UX.

On many UNIX platforms, WL_HOME/common/bin/commEnv.sh script invokes the ulimit -n command to set the upper limit of number of open file descriptors to 1024 before WebLogic Server is started. However, the -n option of the ulimit command is not supported by the ksh shell on HP-UX. So, in WebLogic Server 8.1SP5, the commEnv.sh script does not set the upper limit of open file descriptors for this HP-UX shell.

Operating system: HP-UX 11.0, 11i on PA-RISC

Workaround: You can manually limit the number of open file descriptors. However, this workaround is not available in the ksh shell environment

CR263825

You may experience performance issues due to changes in JRockit SocketMuxer support for JRockit 1.4.2_10 and all later 1.4.2 releases.

Workaround:

Contact your BEA support representative for the CR263825_810sp5 patch.

JDBC Known Issues

Change Request Number

Description

CR251945

When a connection pool failed to satisfy a user request, the multipool tried to disable that pool. However, the disable action could only be done by a user belonging to the "Admin" group. ResourcePermissionsException occurred when the current user did not have the required permissions.

JTA Known Issues

Change Request Number

Description

CR259037

When a transaction times out at logging time, you could experience a deadlock situation. The thread that is trying to write the transaction to the transaction log (TLOG) times out, a different thread trying to execute the rollback to clean the same transaction from the TLOG and deadlock each other.

Workaround or Solution:

Decrease the value of the CheckpointIntervalSeconds attribute and move TLOGs to a directory where they will not run out of memory.

RMI Known Issues

Change Request Number

Description

CR286661

The WebLogic cluster ClusterCommunicationService started before applications were started locally on the server. So the cluster JNDI bindings failed to find certain application-specific classes on the server.

Workaround: Initialize ClusterCommunicationService after all applications are deployed locally on the server.

Security Known Issues

Change Request Number

Description

CR182523

Information on how to configure domains to enable inter-domain transactions (that is, all participating domains run on WebLogic Server 9.x, 8.x, 7.x, and 6.x domains or a combination of 9.x, 8.x, 7.x and 6.x) is incorrect in the Administration Console Help.

Workaround: The information required to configure domains to enable inter-domain transactions is located online. See Configuring Domains for Inter-Domain Transactions in Administration Console Online Help.

Servlets & JSPs Known Issues

Change Request Number

Description

CR262289

When calling the close() method on the InputStream obtained from the HttpsURLConnection class, the underlying SSL socket might not be closed even when the disconnect() method is called. This can result in a socket leak.

Workaround or Solution:

To avoid such a socket leak and ensure all underlying resources are closed, you are advised to use the HttpsURLConnection#disconnect() method instead of the InputStream#close() method.


 

 


Spring Framework on WebLogic Server Issues

Change Request Number

Description and Workaround

CR236708

There is an Antlr conflict between Hibernate 3 and WebLogic Server.

Workaround:

Place Antlr2.7.5.jar before weblogic.jar in your CLASSPATH.

CR237532

There is a WebApp classloading issue with the Spring Framework.

Workaround:

Contact BEA Customer Support for a WebLogic Server-Spring combination patch.

 


Known Issues in WebLogic Server 8.1 SP4

Administration Console Known Issues

Change Request Number

Description

CR186747

When viewing the Administration Console in with the language preference set to Japanese or Korean, the links to customize the information displayed on the page ("Customize this view..." in English) do not work.

Workaround: Select "English" for Language on the Console—>Preferences tab in the Administration Console.

Cluster Known Issues

Change
Request
Number

Description

CR172156

In WebLogic Server 8.1, use of more than about 20 channels in a cluster can result in the formation of multicast header transmissions that exceed the default maximum packet size. The MTUSize attribute in the Server element of config.xml sets the maximum size for packets sent using the associated network card to 1500. Sending packets that exceed the value of MTUSize can result in a java.lang.NegativeArraySizeException, similar to:

java.lang.NegativeArraySizeException at
weblogic.common.internal.WLObjectOutputStreamBase2.writeBytes(WLObjectOutputStreamBase2.java:97)
at weblogic.cluster.MulticastSender.fragmentAndSend(MulticastSender.java:340)
at weblogic.cluster.MulticastSender.send(MulticastSender.java:156) at
weblogic.cluster.AttributeManager.sendAttributes(AttributeManager.java:52) at
weblogic.cluster.ClusterCommunicationService.resume(ClusterCommunicationService.java:79)
at weblogic.t3.srvr.T3Srvr.resume(T3Srvr.java:1042)
at weblogic.t3.srvr.T3Srvr.run(T3Srvr.java:359)
at weblogic.Server.main(Server.java:32)

When the problem was replicated in a cluster with 22 channels, it was solved by setting MTUSize to 2500.

Workaround: If you see NegativeArraySizeExceptions in a cluster with a large number of channels, increase incrementally the value of MTUSize until the exception does not occur.

Core WebLogic Server Known Issues

Change
Request
Number

Description

CR093104

In earlier service packs, the NT Performance pack caused thread deadlock under some circumstances. The problem was exhibited under light loads during stress tests of a JSP accessing a database. Thread dumps showed the majority of "default" execute threads stuck in "waiting for monitor entry":

"ExecuteThread: '10' for queue: 'default'" daemon prio=5 tid=0x273fb548 nid=0xadc waiting for monitor entry [0x2810f000..0x2810fdc4]

Workaround: If you experience such deadlocks, create a .hotspot_compiler file in the directory where you invoke the JVM and put the following lines in that file:

exclude weblogic/socket/NTSocketMuxer processSockets

exclude weblogic/socket/NTSocketMuxer trigger

exclude weblogic/socket/NTSocketMuxer cleanup

CR263825

You may experience performance issues due to changes in JRockit SocketMuxer support for JRockit 1.4.2_10 and all later 1.4.2 releases.

Workaround:

Contact your BEA support representative for the CR263825_810sp4 patch.

CR183137

Certain combinations of Java command-line options can cause the Sun JDK to crash with a HotSpot Virtual Machine Error when JSPs are precompiled in production mode (-server) with debug enabled. The symptom is an error like the following:

# HotSpot Virtual Machine Error, Internal Error
# Please report this error at
# http://java.sun.com/cgi-bin/bugreport.cgi
#
# Java VM: Java HotSpot(TM) Server VM (1.4.2_04-b05 mixed mode)
#
# Error ID: 53484152454432554E54494D450E435050018D
#
# Problematic Thread: prio=5 tid=0x2d48f578 nid=0xc10 runnable

In particular, the following combination of options seems to cause this problem:

  • Domain configured with Sun JDK 1.4.2_04 (not BEA JRockit®)

  • Server running in production mode (-server option)

  • Debug enabled

  • JSP Precompilation enabled (The weblogic.xml file for a Web application specifies precompile=true)

Sun JVM command-line options associated with this issue include:

-server -Xdebug -Xnoagent
-Xrunjdwp:transport=dt_socket,address=8453,server=y,suspend=n
-java.compiler=NONE -ea -da:com.bea... -da:javelin... -da:weblogic...

Workarounds: There are several ways to avoid this issue (choose one):

  • Use the JRockit JVM instead of the Sun JVM.

  • Edit the server start script to start with -client (not -server) when debugging with the Sun JDK.

  • Disable debugging (edit setDomainEnv or manually remove debug JVM arguments).

  • Do not precompile when debugging.

CR239324 CR239819

If a socket is unregistered and then registered, and if between the two calls the timeout timer never runs, the data structures are getting out of sync which prevents the newly-registered socket from ever timing out.

This problem was resolved in 8.1 SP5.

Deployment Known Issues

Change Request Number

Description

CR264560

When the element home-is-clusterable is set to True in weblogic-ejb-jar.xml file and the EJB is deployed in a cluster and if all the cluster members are not started at the same time, there is a possibility of AssertionError and failure to server startup.

Workaround:

  • Precompile the EJB application and verify that the final JAR file has WLStub.class files. If these files do not exist then compile again with -disableHotCodeGen flag.

  • If precompilation is not possible and the EJB is compiled on the server during runtime then add the -disableHotCodeGen flag to the server's extraEJBCOptions. You can add the flag using Server > General Tab > Advanced Options > Extra EJB Compiler Options.

CR210073

When all the modules of an application are pinned (deployed or targeted) to some of the servers in a cluster and no module is targeted to the cluster, a potential deadlock of the Admin RMI threads may cause application deployment to hang.

Workaround: Increase the RMI pool size by specifying -Dweblogic.management.rmiQueueSize=xx in java command in weblogic startup script. It is recommended that the value of xx be equal to one more than the number of managed servers in the domain. The default value for this pool size is 3.

Documentation Known Issues

Change Request Number

Description

CR196210

Previous versions of this document and various other sample documents erroneously described using weblogic.management.Admin.getInstance().getAdminMBeanHome() as a way to look up the MBeanHome interface on the Administration Server.

However, the weblogic.management.Admin class is not public. Instead of using this non-public class, use JNDI to retrieve MBeanHome. See Determining the Active Domain and Servers in Programming WebLogic Server JMX Services.

n/a

Links in the Administration Console Online Help to the JRockit documentation are to the JRockit 8.1 SP3 documentation instead of the JRockit 1.4.2 documentation. A listing of all JRockit documentation can be found at http://download.oracle.com/docs/cd/E13188_01/more_jrockit.html.

Examples Known Issues

Change Request Number

Description

CR175804

The iiop.ejb.entity.tuxclient sample does not run successfully.

This problem is due to changes in JVM behavior regarding vector marshalling. A BEA Tuxedo® 8.1 rolling patch is expected to be released in the near future to resolve this issue.

Installation Known Issues

Change Request Number

Description

CR182698

When upgrading to WebLogic Server 8.1 SP3 on Red Hat Enterprise Linux AS or ES 3.0 on Itanium systems, Smart Update hangs and does not complete the installation. (Smart Update works on Red Hat Enterprise Linux 2.1 on Itanium, but not 3.0.)

Workaround: Do not use Smart Update to upgrade to WebLogic Server 8.1 SP3 on Red Hat Enterprise Linux 3.0. Either use the upgrade installer, or perform a new installation of WebLogic Server 8.1 SP3.

For more information about using the upgrade installer, see "Installing Service Packs and Rolling Patches Using a Downloadable Upgrade Installer" in "Installing Service Packs and Rolling Patches" in Installing BEA WebLogic Platform.

JDBC Known Issues

Change Request Number

Description

CR184298

When using a pooled JDBC connection on a WebLogic Server client, if the JDBC driver classes in the client CLASSPATH are not the exact same version as in the server CLASSPATH, you may see a java.rmi.UnmarshalException. This issue was noted when using a RowSet control in BEA WebLogic Workshop®.

Workaround: Make sure JDBC driver classes in the client and server CLASSPATHs are the same.

CR190393

CountOfTestFailuresTillFlush is not closing all connections when the stipulated number of test failures has been reached, and CountOfRefreshFailuresTillDisable does not disable the pool after the stipulated number of failures.

There is no workaround for this issue in 8.1SP4.

This problem was resolved in 8.1 SP5.

JTA Known Issues

Change Request Number

Description

CR259037

When a transaction times out at logging time, you could experience a deadlock situation. The thread that is trying to write the transaction to the transaction log (TLOG) times out, a different thread trying to execute the rollback to clean the same transaction from the TLOG and deadlock each other.

Workaround or Solution:

Decrease the value of the CheckpointIntervalSeconds attribute and move TLOGs to a directory where they will not run out of memory.

Node Manager Known Issues

Change Request Number

Description

CR203668

When using Managed Servers with the Node Manager, if the Managed Server fails to restart within the time period specified by the Restart Interval attribute and number of attempts specified by the Max Restarts within Interval attribute, the Managed Server will have be restarted manually.

CR209806

When a machine has been configured in the Admin Console and the listen port for the Node Manager is left unspecified, the Admin Server checks the default port 5555 for the presence of a Node Manager. As a result, the Admin Server may encounter conflict with other processes running on port 5555 even when the Node Manager is not deployed.

Security Known Issues

Change Request Number

Description

CR187784

The context path and uri arguments for the URLResource are stored in lowercase on Windows and mixed case on UNIX. As a result, if you define roles and policies in a domain on one operating system, export them, and then import them to a domain on the other operating system, the roles and policies may not work correctly. The end result is that webapp pages that you protect on one operating system may not be protected when the roles and policies are imported to a domain on the other operating system.

CR206178

Using WLS8.1-SP3 with JDK 1.4.2 and making an outbound ssl connection causes the following error:

"java.lang.IllegalStateException: Cipher not initialized ".

The problem is resolved when the customer copied all the jars from the /jre/lib/ext folder of JDK 1.4.1 into the /jre/lib/ext of JDK 1.4.2.

Servlets & JSPs Known Issues

Change Request Number

Description

CR262289

When calling the close() method on the InputStream obtained from the HttpsURLConnection class, the underlying SSL socket might not be closed even when the disconnect() method is called. This can result in a socket leak.

Workaround or Solution:

To avoid such a socket leak and ensure all underlying resources are closed, you are advised to use the HttpsURLConnection#disconnect() method instead of the InputStream#close() method.


 

Transactions Known Issues

Change Request Number

Description

CR201308

If your application is using XA transactions with the Oracle thin driver 10.1.0.3, you must grant execute permission on the dbms_system table to the database user used to connect to the database, along with other database preparations for XA:

1. Execute the following command: @xaview.sql

2. Grant the following permissions:

grant select on v$xatrans$ to public (or <user>);

grant select on pending_trans$ to public;

grant select on dba_2pc_pending to public;

grant select on dba_pending_transactions to public;

(when using the Oracle Thin driver 10.1.0.3 or later)
grant execute on dbms_system to <user>;

If the above steps are not performed on the database server, normal XA database queries and updates may work fine. However, when the Weblogic Server Transaction Manager performs recovery on a re-boot after a crash, recover for the Oracle resource will fail with XAER_RMERR. Crash recovery is a standard operation for an XA resource.

Note: Weblogic Server 8.1SP4 ships with the Oracle Thin driver 10.1.0.2, so granting permission on the dbms_system is required only if you use a different version of the driver.

See "Using the Oracle Thin/XA Driver" in Programming WebLogic JTA for more information about using the Oracle Thin/XA driver with WebLogic Server.

CR220662

Under certain failure conditions following a database server or database network crash, pending transaction branches on failed Oracle Real Application Clusters (RAC) instances may not be properly resolved. Use one of the following methods to correct this problem:

  • To prevent data loss, restart the coordinating managed server

  • If data loss is acceptable, forcibly roll back the transaction in the Oracle database

Web Services Known Issues

Change Request Number

Description

CR107595

SSL does not work for J2ME clients on WebLogic Server 8.1.

Certicom SSL libraries require additional features that are not supported by J2ME. Therefore, SSL is not supported for J2ME clients on WebLogic Server 8.1.

WebLogic Tuxedo Connector (WTC) Known Issues

Change Request Number

Description

CR185475

WTC cannot have multiple connections within a single transaction.

If multiple TuxedoConnections are instantiated within a single transaction, only the first TuxedoConnection instance is enlisted in the transaction.

CR179956

WTC imported service Resource Name must match Remote Name.

WTC imported service Resource Name and Remote Service Name cannot differ. When the imported service Mbean is updated (to give a different resource name), the WTCService cache is not updated.

Workaround: Reboot the WebLogic Server instance.

CR127660

View classes are not set on a per connection basis.

A shared BEA WebLogic Tuxedo ConnectorTM hash table can cause unexpected behavior to the server if two applications point to the same VIEW name with different definitions. There should be a hash table for the view classes on the connection as well as for the Resource section.

Workaround: Ensure that all VIEW classes defined across all your WebLogic Workshop applications are consistent, meaning that you have the same VIEW name representing the same VIEW class.

WebLogic Type 4 JDBC Drivers Known Issues

Change
Request
Number

Description

CR202674

The WebLogic Type 4 JDBC driver for DB2 is not included in the WebLogic Server 8.1 SP4 release. It was removed because DB2 support is not yet available for WebLogic Server 8.1 SP4. For customers that require DB2 support, BEA recommends that you use WebLogic Server 8.1 SP3 until DB2 support is available for WebLogic Server 8.1 SP4.

--

Oracle Driver

The WebLogic JDBC Oracle driver now describes columns defined as FLOAT or FLOAT(n) as a DOUBLE SQL type. The driver previously described these columns as a FLOAT SQL type. Both the DOUBLE type and the FLOAT type represent a double precision floating point number.

--

SQL Server Driver

The SendStringParameterAsUnicode connection option is set to true by default. Setting this option to true can negatively impact performance if an application does string comparison in SQL and the data stored in the database is not Unicode. On the other hand, setting this option to false can cause data corruption or unexpected comparison results if the data is not stored in the default database character set. Microsoft SQL Server allows you to define columns in a table that use a character set other than the default database character set. When sending parameter data to the server, the driver does not know what the character set is for the column that is being compared or modified. Sending parameter data as Unicode will always work, but is subject to the conversion and performance issues mentioned in the Microsoft document: http://support.microsoft.com/default.aspx?scid=kb;en-us;Q271566.

CR183190

Under certain statement failure conditions, cached statements are leaked without being closed, which can lead to DBMS resource problems. When using a DB2 database, you may see the following error:

"java.sql.SQLException: [BEA][DB2 JDBC Driver]No more available statements.

Please recreate your package with a larger dynamicSections value."

There is no workaround for this issue in 8.1SP4.

This problem was resolved in 8.1 SP5.

 


Known Issues in WebLogic Server 8.1 SP3

The following sections describe known issues in WebLogic Server 8.1 SP3:

Administration Console Known Issues

Change Request Number

Description

CR186747

When viewing the Administration Console in with the language preference set to Japanese or Korean, the links to customize the information displayed on the page ("Customize this view..." in English) do not work.

Workaround: Select "English" for Language on the Console—>Preferences tab in the Administration Console.

Cluster Known Issues

Change
Request
Number

Description

CR172156

In WebLogic Server 8.1, use of more than about 20 channels in a cluster can result in the formation of multicast header transmissions that exceed the default maximum packet size. The MTUSize attribute in the Server element of config.xml sets the maximum size for packets sent using the associated network card to 1500. Sending packets that exceed the value of MTUSize can result in a java.lang.NegativeArraySizeException, similar to:

java.lang.NegativeArraySizeException at
weblogic.common.internal.WLObjectOutputStreamBase2.writeBytes(WLObjectOutputStreamBase2.java:97)
at weblogic.cluster.MulticastSender.fragmentAndSend(MulticastSender.java:340)
at weblogic.cluster.MulticastSender.send(MulticastSender.java:156) at
weblogic.cluster.AttributeManager.sendAttributes(AttributeManager.java:52) at
weblogic.cluster.ClusterCommunicationService.resume(ClusterCommunicationService.java:79)
at weblogic.t3.srvr.T3Srvr.resume(T3Srvr.java:1042)
at weblogic.t3.srvr.T3Srvr.run(T3Srvr.java:359)
at weblogic.Server.main(Server.java:32)

When the problem was replicated in a cluster with 22 channels, it was solved by setting MTUSize to 2500.

Workaround: If you see NegativeArraySizeExceptions in a cluster with a large number of channels, increase incrementally the value of MTUSize until the exception does not occur.

Core WebLogic Server Known Issues

Change
Request
Number

Description

CR093104

In earlier service packs, the NT Performance pack caused thread deadlock under some circumstances. The problem was exhibited under light loads during stress tests of a JSP accessing a database. Thread dumps showed the majority of "default" execute threads stuck in "waiting for monitor entry":

"ExecuteThread: '10' for queue: 'default'" daemon prio=5 tid=0x273fb548 nid=0xadc waiting for monitor entry [0x2810f000..0x2810fdc4]

Workaround: If you experience such deadlocks, create a .hotspot_compiler file in the directory where you invoke the JVM and put the following line in that file:

exclude weblogic/socket/NTSocketMuxer processSockets

CR263825

You may experience performance issues due to changes in JRockit SocketMuxer support for JRockit 1.4.2_10 and all later 1.4.2 releases.

Workaround:

Contact your BEA support representative for the CR263825_810sp3 patch.

CR183137

Certain combinations of Java command-line options can cause the Sun JDK to crash with a HotSpot Virtual Machine Error when JSPs are precompiled in production mode (-server) with debug enabled. The symptom is an error like the following:

# HotSpot Virtual Machine Error, Internal Error
# Please report this error at
# http://java.sun.com/cgi-bin/bugreport.cgi
#
# Java VM: Java HotSpot(TM) Server VM (1.4.2_04-b05 mixed mode)
#
# Error ID: 53484152454432554E54494D450E435050018D
#
# Problematic Thread: prio=5 tid=0x2d48f578 nid=0xc10 runnable

In particular, the following combination of options seems to cause this problem:

  • Domain configured with Sun JDK 1.4.2_04 (not JRockit)

  • Server running in production mode (-server option)

  • Debug enabled

  • JSP Precompilation enabled (The weblogic.xml file for a Web application specifies precompile=true)

Sun JVM command-line options associated with this issue include:

-server -Xdebug -Xnoagent
-Xrunjdwp:transport=dt_socket,address=8453,server=y,suspend=n
-java.compiler=NONE -ea -da:com.bea... -da:javelin... -da:weblogic...

Workarounds: There are several ways to avoid this issue (choose one):

  • Use the JRockit JVM instead of the Sun JVM.

  • Edit the server start script to start with -client (not -server) when debugging with the Sun JDK.

  • Disable debugging (edit setDomainEnv or manually remove debug JVM arguments).

  • Do not precompile when debugging.

CR240499

In HP-UX, the thread runs with the default thread-priority assigned by the operating system even when a different thread priority is set from the Administration Console.

Workaround:

    1. As a root user, run the following command

setprivgrp <user-group> RTSCHED

where, user-group is the unix-group.

    2. Add the following java command line option to start WebLogic script

-XX:SchedulerPriorityRange=SCHED_RTPRIO

Documentation Known Issues

Change Request Number

Description

n/a

Links in the Administration Console Online Help to the JRockit documentation are to the JRockit 8.1 SP3 documentation instead of the JRockit 1.4.2 documentation. A listing of all JRockit documentation can be found at http://download.oracle.com/docs/cd/E13188_01/more_jrockit.html.

Examples Known Issues

Change Request Number

Description

CR175804

The iiop.ejb.entity.tuxclient sample does not run successfully.

This problem is due to changes in JVM behavior regarding vector marshalling. A Tuxedo 8.1 rolling patch is expected to be released in the near future to resolve this issue.

Installation Known Issues

Change Request Number

Description

CR182698

When upgrading to WebLogic Server 8.1 SP3 on Red Hat Enterprise Linux AS or ES 3.0 on Itanium systems, Smart Update hangs and does not complete the installation. (Smart Update works on Red Hat Enterprise Linux 2.1 on Itanium, but not 3.0.)

Workaround: Do not use Smart Update to upgrade to WebLogic Server 8.1 SP3 on Red Hat Enterprise Linux 3.0. Either use the upgrade installer, or perform a new installation of WebLogic Server 8.1 SP3.

For more information about using the upgrade installer, see "Installing Service Packs and Rolling Patches Using a Downloadable Upgrade Installer" in "Installing Service Packs and Rolling Patches" in Installing BEA WebLogic Platform.

JDBC Known Issues

Change Request Number

Description

CR179600

Under certain statement failure conditions, cached statements are leaked without being closed, which can lead to DBMS resource problems. When using an Oracle database, this can result in out-of-cursor failures.

A patch is available for this issue. To get the patch, contact BEA Customer Support and reference CR179600.

This problem was resolved in 8.1 SP4.

CR183190

Under certain statement failure conditions, cached statements are leaked without being closed, which can lead to DBMS resource problems. When using a DB2 database, you may see the following error:

"java.sql.SQLException: [BEA][DB2 JDBC Driver]No more available statements.

Please recreate your package with a larger dynamicSections value."

There is no workaround for this issue in 8.1SP3.

This problem was resolved in 8.1 SP5.

CR184298

When using a pooled JDBC connection on a WebLogic Server client, if the JDBC driver classes in the client CLASSPATH are not the exact same version as in the server CLASSPATH, you may see a java.rmi.UnmarshalException. This issue was noted when using a RowSet control in WebLogic Workshop.

Workaround: Make sure JDBC driver classes in the client and server CLASSPATHs are the same.

CR190393

CountOfTestFailuresTillFlush is not closing all connections when the stipulated number of test failures has been reached, and CountOfRefreshFailuresTillDisable does not disable the pool after the stipulated number of failures.

There is no workaround for this issue in 8.1SP3.

This problem was resolved in 8.1 SP5.

JTA Known Issues

Change Request Number

Description

CR259037

When a transaction times out at logging time, you could experience a deadlock situation. The thread that is trying to write the transaction to the transaction log (TLOG) times out, a different thread trying to execute the rollback to clean the same transaction from the TLOG and deadlock each other.

Workaround or Solution:

Decrease the value of the CheckpointIntervalSeconds attribute and move TLOGs to a directory where they will not run out of memory.

RMI/IIOP Known Issues

Change Request Number

Description

CR124596
CR192252

The enhancement described in CR124596 was not fully implemented in WebLogic Server 8.1SP3. The optional enhancement to the BEA ORB forces reconnection when bootstrapping and allows hardware load-balancers to correctly balance connection attempts.

A patch for WebLogic Server 8.1SP3 is now available. To get the patch, contact BEA Customer Support and reference the patch for CR192252.

This problem was resolved in 8.1 SP4 for CR192252.

Security Known Issues

Change Request Number

Description

CR184421

When importing security data into a new security realm, the user weblogic loses membership in all groups except Administrators. After importing the data, you must manually add weblogic to any required groups.

CR187784

The context path and uri arguments for the URLResource are stored in lowercase on Windows and mixed case on UNIX. As a result, if you define roles and policies in a domain on one operating system, export them, and then import them to a domain on the other operating system, the roles and policies may not work correctly. The end result is that webapp pages that you protect on one operating system may not be protected when the roles and policies are imported to a domain on the other operating system.

Servlets & JSPs Known Issues

Change Request Number

Description

CR262289

When calling the close() method on the InputStream obtained from the HttpsURLConnection class, the underlying SSL socket might not be closed even when the disconnect() method is called. This can result in a socket leak.

Workaround or Solution:

To avoid such a socket leak and ensure all underlying resources are closed, you are advised to use the HttpsURLConnection#disconnect() method instead of the InputStream#close() method.


 

Transactions Known Issues

Change Request Number

Description

CR193458

When using DataDirect Connect for BEA WebLogic JDBC DriversTM version 3.4 with a DB2 database, you may see XAER_RMERR errors for applications that previously ran without errors using the 3.3 version of the drivers. The problem is caused by implicit closing of result sets, a feature added to the drivers to improve performance.

Workaround: DataDirect added a connection property to control implicit result set closing: AllowImplicitResultSetCloseForXA. The feature is enabled by default. To disable implicit result set closings, set AllowImplicitResultSetCloseForXA=false. To make this setting for a JDBC connection pool, add it to the Properties for the connection pool on the JDBC Connection Pool —> Configuration —> General tab in the Administration Console. The resulting change in the config.xml file resembles the following config.xml excerpt:

<JDBCConnectionPool
DriverName="com.ddtek.jdbcx.db2.DB2DataSource"
Name="myPool"
Properties="user=scott;
portNumber=50000;
databaseName=SAMPLE;
serverName=dbserver1.yourco.COM;
batchPerformanceWorkaround=true;
AllowImplicitResultSetCloseForXA=false"
...

URL="jdbc:datadirect:db2://dbserver1:50000;DatabaseName=SAMPLE" />

Note: This issue applies to the DataDirect Connect for JDBC version 3.4 drivers for DB2 only. The problem has not been observed with drivers for other database management systems or previous releases of the DB2 drivers. AllowImplicitResultSetCloseForXA is ignored for these other drivers.

Web Services Known Issues

Change Request Number

Description

CR107595

SSL does not work for J2ME clients on WebLogic Server 8.1.

Certicom SSL libraries require additional features that are not supported by J2ME. Therefore, SSL is not supported for J2ME clients on WebLogic Server 8.1.

WebLogic Tuxedo Connector (WTC) Known Issues

Change Request Number

Description

CR185475

WTC cannot have multiple connections within a single transaction.

If multiple TuxedoConnections are instantiated within a single transaction, only the first TuxedoConnection instance is enlisted in the transaction.

CR179956

WTC imported service Resource Name must match Remote Name.

WTC imported service Resource Name and Remote Service Name cannot differ. When the imported service Mbean is updated (to give a different resource name), the WTCService cache is not updated.

Workaround: Reboot the WebLogic Server instance.

CR127660

View classes are not set on a per connection basis.

A shared WebLogic Tuxedo Connector hash table can cause unexpected behavior to the server if two applications point to the same VIEW name with different definitions. There should be a hash table for the view classes on the connection as well as for the Resource section.

Workaround: Ensure that all VIEW classes defined across all your WebLogic Workshop applications are consistent, meaning that you have the same VIEW name representing the same VIEW class.


 

 


Known Issues in WebLogic Server 8.1 SP2

The following sections describe known issues in WebLogic Server 8.1 SP2:

Patch Available for Sun Java 2 1.4.2 SDK, Oracle 10g Driver, and SQL Server, Sybase or DB2 Database Users

If you are using WebLogic Server or BEA WebLogic PlatformTM 8.1 SP2 with the Sun Java 2 1.4.2 SDK, or the Oracle 10g driver, or a SQL Server, Sybase or DB2 database, you may need to install the WebLogic Platform 8.1 SP2 SDK1.4.2/Oracle10gdriver/Database patch. This patch, and a description of the specific configurations that require it, are available at the following dev2dev Web site:

http://dev2dev.bea.com/products/wlplatform81/patch/wlplat81sp2_patch.jsp

Administration Console Known Issues

Change Request Number

Description

CR125552

When using the WebLogic Server 8.1 Administration Console on machines that have JDK version 1.4.1_02 installed, the navigation applet fails to load and the system freezes.

Workaround: Uninstall JDK version 1.4.1_02 from the system and reinstall version JDK version 1.4.2_01 or later.

CR129594

The WebLogic Server Administration Console's JDBC Connection Pool Configuration tab does not have an option to rollback local transactions upon closing a connection.

Workaround: Set this option by editing the config.xml JDBCConnectionPool attribute RollbackLocalTxUponConnClose, or via the Configuration Wizard.

CR132229

The separators for multiple addresses in BEA JoltTM and WLEC ConnectionPools are incorrectly identified as semicolons.

Workaround: Use commas to separate multiple addresses when configuring Jolt and WLEC Primary and Secondary Address fields.

Configuration Wizard Known Issues

Change Request Number

Description

CR125514

When running the config_builder.sh script from the BEA_HOME/wls81/common/bin directory on a Solaris platform, the Configuration Template Builder does not start. The message "Unable to instantiate GUI, defaulting to console mode" is displayed, but rather than starting up in Console mode, a command prompt is displayed. The Configuration Template Builder is not supported in console mode—a graphics environment is required.

Workaround: Run the Configuration Wizard using xWindows.

Note: The error message will be changed in the next service pack to more accurately communicate this limitation.

Connector Known Issues

Change Request Number

Description

n/a

When a connection request is made, WebLogic Server returns to the client (via the resource adapter) a Proxy object that wraps the connection object. WebLogic Server uses this proxy to provide features that assist applications using the WebLogic Server implementation of the J2EE Connector Architecture. These features include (1) connection leak detection capabilities and (2) late XAResource enlistment when a connection request is made before starting a global transaction that uses that connection.

If the connection object returned from a connection request is cast as a Connection class, a ClassCastException can occur. This exception is caused by either (1) the resource adapter doing the cast or (2) the client doing the cast during a connection request.

In WebLogic Server 8.1 SP2, an attempt is made to detect the ClassCastException caused by the resource adapter case (1) above. If the server detects that this cast is failing, it turns off the proxy wrapper feature and proceeds by returning the connection object during a connection request (unwrapped). The server logs a warning message to indicate that the proxy wrapper has been turned off. When this type of cast failure occurs, connection leak detection and late XAResource enlistment features are also turned off (but currently no indication of this is given in the Console monitoring).

WebLogic Server attempts to detect the ClassCastException by acting as a client using container-managed security. Doing so requires the resource adapter to be deployed with security credentials defined.

If the client is doing the cast and getting a ClassCastException, the customer (client) code can be modified as follows:

Workaround: If the client casts the connection object to MyConnection, instead of MyConnection being a class that implements the resource adapter's Connection interface, modify it to be an interface that extends Connection. Implement a MyConnectionImpl class that implements the MyConnection interface.

CR100048

If two adapters are configured to use the same Enterprise Information System (EIS) with the same Resource Manager, then using the two adapters in the same transaction can result in deadlocks or long delays followed by a transaction failure. This is because most Resource Managers, including Oracle's, do not allow adapters to share an EIS.

CR127127

The IIOP thin-client is not supported on AIX (or other platforms with the IBM VM) due to dependencies on the Sun VM.

Note: Only clients are affected by this issue. A thin-client running on another platform will work just fine against WebLogic Server running on AIX.

Core Known Issues

Change Request Number

Description

CR106616

After several operations on the WebLogic Server Administration Console, Netscape Version 4.79 running on Linux AS2.1 and HP-UX shuts down with a bus error, sometimes with a core dump.

CR263825

You may experience performance issues due to changes in JRockit SocketMuxer support for JRockit 1.4.2_10 and all later 1.4.2 releases.

Workaround:

Contact your BEA support representative for the CR263825_810sp2 patch.

EJB Known Issues

Change Request Number

Description

CR112393

An EJB JAR is referencing another utility JAR through the Manifest classpath. When you compile an EJB using the weblogic.appc utility, classes included in the referenced JAR are not found. The appc compiler explodes the source JAR in a temporary directory and the classfinder looks for the dependent JAR in that temporary directory; however, the classes are not copied into the temporary directory.

Workaround:

This issue only exists for stand-alone modules that are not part of an EAR file. BEA recommends that you package the EJB JAR file and utility JAR file together in an EAR file. This will ensure they are always tied together. If you choose not to do this, BEA recommends you use the weblogic.ejbc utility rather than weblogic.appc.

Note: The weblogic.ejbc utility is a deprecated utility and will not be available in future releases.

CR126613

An MDB running in a transactional context interacts with an external system (such as an LDAP server) and message processing takes longer than transaction timeout value. The MDB throws the following exception and message processing stops:

weblogic.utils.AssertionError: ***** ASSERTION FAILED *****[ Left-over JTA transaction found on MDB listener thread ] at
weblogic.utils.Debug.assertion(Debug.java:57) at
weblogic.ejb20.internal.MDListener.onMessage(MDListener.java:245) at
weblogic.jms.client.JMSSession.onMessage(JMSSession.java:2596) at
weblogic.jms.client.JMSSession.execute(JMSSession.java:2516) at
weblogic.kernel.ExecuteThread.execute(ExecuteThread.java:197) at
weblogic.kernel.ExecuteThread.run(ExecuteThread.java:170)

Workaround: Increase the value of the transaction timeout to make it high enough.

Examples Known Issues

Change Request Number

Description

CR104550

The config.xml file for the samples banking application domain contains an incorrect name for the EJB. With this error, the application cannot find the EJB.

<Application Deployed="true" Name="app_banking"
 Path=".\config\migrationdomain\applications"
 StagedTargets="migrationserver">
<EJBComponent Name="app_banking"
 Targets="migrationserver"
 URI="app_banking.jar"/>
</Application>

It should contain:

<Application Deployed="true"
 Name="containerManaged"
 Path=".\config\migrationdomain\applications"
 StagedTargets="migrationserver">
<EJBComponent Name="containerManaged"
 Targets="migrationserver"
 URI="app_banking.jar"/>
</Application>

Workaround: Undeploy the EJB, and redeploy it with the correct name, containerManaged.

General Known Issues

Change Request Number

Description

CR128912

There is a known issue in HP-UX SDK 1.4.1.05 character set API and as a result, Charset.availableCharsets() always fails. This problem manifests itself in WebLogic Server SP2 using JVM 1.4.1.05.

Workaround: Download and use HP's 1.4.1.06 JVM, available as part of the JDK 1.4.1.06 at http://www.hp.com/products1/unix/java/.

Installation Known Issues

Change Request Number

Description

CR127350

As part of the upgrade procedure, the WebLogic Platform 8.1 SP2 installation program automatically overwrites any files, including applications, that you have created in WL_HOME.

Workaround: Before installing SP2, make backup copies of any files or applications that you have created in WL_HOME.

JMS Known Issues

Change Request Number

Description

CR125979

Customers who set up a JMS bridge within a cluster may find that when starting the bridge, the following exception is thrown from the Managed Servers, and that the JMS message cannot be sent to the intended destination.

<Oct 16, 2003 3:12:49 PM PDT> <Notice> <WebLogicServer> <BEA-000360>
<Server started in RUNNING mode>
<Oct 16, 2003 3:23:17 PM PDT> <Error> <JMS> <BEA-040368> <The following exception has occurred: java.lang.NullPointerException at weblogic.jms.bridge.internal.MessagingBridge.startInternal(MessagingBridge.java:519) at
weblogic.jms.bridge.internal.MessagingBridge.execute(MessagingBridge.java:956) at
weblogic.kernel.ExecuteThread.execute(ExecuteThread.java:197) at
weblogic.kernel.Kernel.execute(Kernel.java:336) at
weblogic.kernel.Kernel.execute(Kernel.java:321) at
...

Workaround:

Deploy the adapter to individual servers instead of to the entire cluster. The adapter must be deployed on the same server to which the bridge is targeted. For instructions, see http://download.oracle.com/docs/cd/E13222_01/wls/docs81/ConsoleHelp/messaging_bridge.html#1109910.

This problem was resolved in 8.1 SP3.

CR128596

In some cases, messages remain in the source queue of the JMS bridge in the sending-side cluster after restarting a Managed Server on the receiving side cluster.

Workaround: Apply the patch for CR128596 to all WebLogic Server instances.

This problem was resolved in 8.1 SP3.

JTA Known Issues

Change Request Number

Description

CR259037

When a transaction times out at logging time, you could experience a deadlock situation. The thread that is trying to write the transaction to the transaction log (TLOG) times out, a different thread trying to execute the rollback to clean the same transaction from the TLOG and deadlock each other.

Workaround or Solution:

Decrease the value of the CheckpointIntervalSeconds attribute and move TLOGs to a directory where they will not run out of memory.

JVM Known Issues

Change Request Number

Description

CR131717

Due to a bug in the Sun JDK version 1.4.2_02, a server ExecuteThread freezes or a client JVM hangs when using JDK version 1.4.2_02 with Japanese encodings ISO-2022-JP and ISO2022JP. Customers not using Japanese encodings will not be affected by this bug.

Workaround: This bug will be fixed in JDK version 1.4.2_04 (available February 12, 2004), as documented at http://developer.java.sun.com/developer/bugParade/bugs/4879522.html. In the meantime, Japanese customers can use JDK version 1.4.1.

CR132292

A change in behavior of javac in JDK version 1.4.2 causes tag handler classes written as inner classes to generate compilation errors. The code generated contains a "$" character in the type name, which is not allowed in JDK version 1.4.2.

Workaround: Declare tag handler classes as public classes instead of inner classes. Or, contact support to obtain a patch for WebLogic Server 8.1 SP2.

CR135783

Calling WebLogicMBeanMaker to package authentication provider MBeans yields different results between JDK version 1.4.1_05 and 1.4.2_03.

With JDK 1.4.1_05, when calling WebLogicMBeanMaker with a source folder that contains no class files, WebLogicMBeanMaker silently ignores the error and creates an MBean JAR file.

With JDK 1.4.2_03, WebLogicMBeanMaker fails if there are no class files in the source path.

CR136167

The Administration Console cannot display the server log in a Japanese locale because of a problem in JDK 1.4.2_03 and earlier JDK 1.4.2 releases.

Workaround: Use the Sun JDK1.4.2_04 or JRockit JDK1.4.2_04.

Plug-In Known Issues

Change Request Number

Description

CR130088

The IIS proxy plugin performs very slowly when the IIS HTTP Keep-Alives Enabled check box is checked. The symptom is a 20-40x slowdown compared to normal performance. It is also possible to receive PROTOCOL_ERROR messages under certain keepalive configurations.

Workaround: If you see slow performance or PROTOCOL_ERROR messages with the IIS proxy plugin, switch to the following IIS/proxy plugin keepalive configuration:

    1. Configure IIS so the 'HTTP Keep-Alives Enabled' check box is NOT checked.

Open the Internet Services Manager tool so you can configure IIS.

In the left-hand pane, right-click on the Web site item which proxies activity for your server.

Select 'Properties' Select the 'Web Site' tab.

Ensure the 'HTTP Keep-Alives Enabled' check box is NOT checked.

Click Apply, then OK.

    2. Configure the proxy plugin so KeepAliveEnabled is false in the iis_proxy.ini file.

Edit the iis_proxy.ini file in a text editor, and set KeepAliveEnabled=false.

    3. Restart IIS.

In the left-hand pane, right-click on the IIS server name, and select Restart IIS...

Select Restart Internet Services on <IIS servername>.

Click OK.

Once IIS restarts, the proxy performance should improve to normal levels.

This problem was resolved in 8.1 SP3.

Security Known Issues

Change Request Number

Description

CR064593

In the new security framework in WebLogic Server, SSL does not cache the sessions; instead SSL initiates a new handshake for each request causing performance issues and out of memory exceptions.

In WebLogic 7.0 and 8.1 SP2, WebLogic Server added the weblogic.security.SSL.sessionCache.ttl which maintains the time to live for the SSL session. The default value is 90000 milliseconds (90 seconds) which means if a client accesses the server again (via the same session ID) within 90 seconds, WebLogic Server will use the existing SSL session. You can change this value by setting -Dweblogic.security.SSL.sessionCache.ttl in the server startup script.

CR126610

Customers using a BEA WebLogic ExpressTM license are not able to configure global security roles or global security policies using the WebLogic Server Administration Console. This behavior is as designed, and this limitation will be updated in the documentation.

CR130185

The JDK cacerts file and the WebLogic Server cacerts file are keystore files that contain out-of-the-box trusted Certificate Authority certificates. These trusted CA certs are used in certificate chain verification when using SSL.

All certificates, including trusted CAs, have NotBefore and NotAfter dates contained within them. A certificate cannot be validated prior to or after these validity dates, respectively.

In this service pack, WebLogic Server checks each trusted CA certificate in the JDK and WebLogic Server cacerts files for the NotAfter date. If the trusted CA is due to expire (that is, exceed the NotAfter date) within 30 days, a warning message is output about this impending expiration. If a trusted CA has already expired, with one exception, a warning message is output declaring the trusted CAs as having exceeded the NotAfter date.

There are two trusted CAs in the JDK cacerts keystore file due to expire approximately at the release date of this service pack. Thus, you may see one or both of these warning messages output while booting WebLogic Server.

Workaround: If you use either of these two trusted CAs in your server certificate chain, you should obtain a new trusted CA or certificate from your Certificate Authority of choice (Verisign, Certicom, and so on). For more information about expiring and renewing certificates, contact your preferred Certificate Authority.

The current JDK cacerts file contains at least one certificate from Verisign that has expired (the Class 4 Certificate Authority, alias verisignclass4ca). No certificate expired message is output when this trusted CA is loaded.

For more information, see Sun Alert 57436 at http://sunsolve.sun.com/pub-cgi/retrieve.pl?doc=fsalert%2F57436.

Servlet & JSP Known Issues

Change Request Number

Description

CR125201

A ClassNotFoundException occurs when migrating sessions to WebLogic Server 8.1 SP2 using secure URLs.

Note: WebLogic Server does not support two servers in different releases interacting in this way.

CR126809

The Sessions High count for Web Applications is not accurate, and only displays counts from the time the server was started.

Note: The Open Sessions count, which previously had this problem, was recently fixed and is accurate.

CR172752

During Web application undeployment, a servlet will throw NPEs (NullPointerExceptions). This is a Sun servlet specification limitation. To avoid this, the application code should call "super.init(ServletConfig)" in the servlet init() call.

Note: This a known limitation.

CR262289

When calling the close() method on the InputStream obtained from the HttpsURLConnection class, the underlying SSL socket might not be closed even when the disconnect() method is called. This can result in a socket leak.

Workaround or Solution:

To avoid such a socket leak and ensure all underlying resources are closed, you are advised to use the HttpsURLConnection#disconnect() method instead of the InputStream#close() method.

Transactions Known Issues

Change Request Number

Description

CR130269

By default, database connections that use an XA driver have autoCommit set to false. If you use a database connection that is based on an XA driver for a read-only operation, the connection implicitly creates a local transaction. When the connection is returned to the connection pool, the local transaction still exists on the connection. The next time the connection is used, WebLogic Server internally calls XAResource.start(), which fails because there is already a transaction on the connection.

Workaround: Set autoCommit to true or call connection.commit() before returning the connection to the connection pool.

WebLogic JDBC Type 4 Drivers Known Issues

Change Request Number

Description

n/a

DB2 Driver

  • Because of lack of support in the DRDA listener for all platforms except DB2 UDB v8.1, ResultSetMetaData describes BIGINT columns as DECIMAL, and stored procedures accepting BIGINT as an input parameter fail on execution with a conversion error.

  • Scroll-sensitive result sets are not supported. Requests for scroll-sensitive result sets are downgraded to scroll-insensitive result sets when possible. When this happens, a warning is generated indicating that result set scrollability has been downgraded.

  • The WebLogic Type 4 JDBC DB2 driver must be able to determine the data type of the column or stored procedure argument to implicitly convert the parameter value. Not all DB2 database versions support getting parameter metadata for prepared statements. Implicit conversions are not supported on database versions that do not provide parameter metadata for prepared statements.

n/a

DB2 Driver on Windows and UNIX

  • Except for DB2 UDB 8.1, CLOB data types are limited to 32 K characters because of DRDA listener limitations.

  • Except for DB2 UDB 8.1, BLOB data types are not supported because of DRDA listener limitations.

n/a

DB2 Driver on iSeries and AS/400

  • Except for DB2 iSeries V5R2, BLOB and CLOB data types are not supported.

n/a

MS SQL Server Driver

  • To ensure correct handling of character parameters, install Service Pack 2, or higher, for SQL Server 7 installations.

CR111002

When using a configuration that has more than one connection pool that uses the WebLogic Type 4 JDBC driver for MS SQL Server, it is possible that WebLogic Server may hang during transaction recovery. This is due to a problem in Microsoft Distributed Transaction Coordinator (DTC) and SQL Server.

Until a hotfix is publicly available from Microsoft, call BEA Support.

n/a

Oracle Driver

  • If you want to take advantage of JDBC distributed transactions through JTA with the Oracle driver, you must install Oracle version 8.1.7 or higher, and install the Oracle JAVA_XA package on the database server.

  • Because JDBC does not support a cursor data type, the Oracle driver returns refcursor output parameters to the application as result sets. Your application should not include a parameter marker or should not declare the refcursor output parameter. The driver transparently converts the refcursor to a result set, which can be retrieved using getResultSet or getMoreResults.

  • By default, values for TIMESTAMP WITH TIME ZONE columns cannot be retrieved using the ResultSet.getTimestamp method because the time zone information would be lost. The driver returns NULL when the getTimestamp method is called on a TIMESTAMP WITH TIME ZONE column. It generates a SQLException in that case. For details about using the TIMESTAMP WITH TIME ZONE data type with the Oracle driver, see TIMESTAMP WITH TIME ZONE Data Type in WebLogic Type 4 JDBC Drivers.

CR120330

Sybase Driver

The getDatabaseProductName method returns "SQL Server," which is the string returned internally by the Sybase database server. This value may not be the same return as seen with other JDBC drivers, including the Sybase JConnect JDBC drivers.

Workaround: If you need the return string to include "Sybase," use the getDriverName method.

CR123475

Sybase XA Driver

When using the Sybase XA driver connecting to a Sybase 12.5.0.3 database and calling getMetaData() after some DDL operations, the method erroneously returns null.

Workaround:

In the connection properties, set useAlternateMetaData=true. For example:

In a connection pool, the configuration may look like:

<JDBCConnectionPool DriverName="weblogic.jdbcx.sybase.SybaseDataSource" Name="MyJDBCConnectionPool" Password="{3DES}x1KgkDjghuQ=" Properties="user=scott;url=jdbc:bea:sybase://sybserver:4100; useAlternateMetaData=true;PortNumber=4100;ServerName=sybserver;DatabaseName=wl" Targets="mycluster" URL="jdbc:bea:sybase ://sybserver:4100"/>

In a non-pooled connection, code may look like:

SybaseDataSource ds = new SybaseDataSource();
ds.setUser("USER");
ds.setPassword("PASSWORD");
ds.setServerName("SERVER");
ds.setPortNumber(5000);
ds.setExtendedOptions("useAlternateMetaData=true");
xaConn = ds.getXAConnection();
conn = xaConn.getConnection();

n/a

All Drivers

  • The WebLogic Type 4 JDBC drivers support retrieval of output parameters from a stored procedure before all result sets and/or update counts have been completely processed. When CallableStatement.getXXX is called, result sets and update counts that have not yet been processed by the application are discarded to make the output parameter data available. Warnings are generated when results are discarded.

  • The WebLogic Type 4 JDBC drivers allow PreparedStatement.setXXX and ResultSet.getXXX methods on BLOB/CLOB data types, in addition to what is described in the JDBC specification. The supported conversions are generally the same as those for LONGVARBINARY/LONGVARCHAR, except where limited by database support.

  • Calling CallableStatement.registerOutputParameter(parameterIndex, sqlType) with sqlType Types.NUMERIC or Types.DECIMAL sets the scale of the output parameter to zero. According to the JDBC specification, calling CallableStatement.registerOutputParameter(parameterIndex, sqlType, scale) is the recommended method for registering NUMERIC or DECIMAL output parameters.

  • When attempting to create an updatable, scroll-sensitive result set for a query that contains an expression as one of the columns, the driver cannot satisfy the scroll-sensitive request. The driver downgrades the type of the result returned to scroll-insensitive.

  • The preferred method for executing a stored procedure that generates result sets and update counts is using CallableStatement.execute(). If multiple results are generated using executeUpdate, the first update count is returned. Any result sets prior to the first update count are discarded. If multiple results are generated using executeQuery, the first result set is returned. Any update counts prior to the first result set are discarded. Warnings are generated when result sets or update counts are discarded.

  • The result set methods getTimestamp, getDate, and getTime return references to mutable objects. If the object reference returned from one of these methods is modified, then re-fetching the column using the same method will return the modified value. The value is only modified in memory; the database value is not modified.

CR124744

All Drivers

When calling Statement.setQueryTimeout(), SQL statements take twice as long as specified to time out.

Workaround: Set the query timeout to half as long as desired.

This problem was resolved in 8.1 SP3.

CR178619

XA Driver for DB2 and Microsoft SQL Server

Drivers return the error code of RMERR (-3) instead of NOTA (-4) which causes transactions to be retried until AbandonTimoutSeconds instead of the transaction being resolved on recovery.

Web Services Known Issues

Change Request Number

Description

CR127610

When invoking a WebLogic Web Service operation using the stubs generated by the clientgen Ant task, the client application might sometimes throw a ClassCastException error if the method that implements the operation uses a global element of a global complex type. The problem is that the clientgen Ant task, when generating the stubs, uses the wrong parameter class due to the data type mapping information in the provided types.xml file being overwritten by JAX-RPC default.

CR128123

A WebLogic Server 8.1 clientgen-generated Web Service that uses SSL transport is not supported on a target client platform based on a JVM version below JDK 1.4.1.

CR131690

When sending a SOAP request that contains encrypted multi-byte strings to the Web Service, the Web Service client cannot specify Shift_JIS or EUC-JP into the message's encoding weblogic.webservice.i18n.charset. If those encodings are specified, the Web Service returns a SOAPFaultException. When using UTF-8, this issue does not occur.

Workaround: Do not set Shift_JIS and EUC-JP in weblogic.webservice.i18n.charset when sending encrypted multi-byte strings. Set UTF-8 instead.

CR175036

If a WSDL has elements with the same name but with a difference in the case (upper or lower) of one or more letters, running clientgen on the WSDL fails with an error such as:

C:\mydir\resources>java weblogic.webservice.clientgen -packageName cargoPackage -wsdl

WSDLName.wsdl -clientJar case.jar C:\DOCUME~1\anurag\LOCALS~1\Temp\case.jar-1007802369\org\openuri\www\FooCodec.java:14:

class FOOCode

c is public, should be declared in a file named FOOCodec.java public final class FOOCodec

This is a system limitation of Windows 32-bit operating systems. You cannot generate FOO.java and Foo.java in the same Windows (32-bit) directory because the operating system is case-insensitive. The second file will always overwrite the first one.

 


Known Issues in WebLogic Server 8.1 SP1

The following sections describe known issues in WebLogic Server 8.1 SP1:

Administration Console Known Issues

Change Request Number

Description

CR111676

The navigation tree in the left pane of the WebLogic Server Administration Console will not display if it is running directly on a system with the following configurations:

  • Windows 64-bit platform with Internet Explorer 6

  • Linux 64-bit platform with Mozilla 1.0.1

Workarounds:

  • Obtain the Java plug-in from the operating system vendor's Web site. If the plug-in is not available, run the Administration Console on a system with a configuration that supports the Java plug-in.

  • Run the Administration Console on a system with a different configuration.

Cluster Known Issues

Change Request Number

Description

CR105861

When a Web application and EJB are deployed in two different clusters and an EJB stub is placed into the HTTP session, a java.lang.ClassCastException is thrown.

Workaround:

Rather than putting the stub into the HTTP session, you should put the Handle into the HTTP session and also ensure that any custom application classes are in the system classpath. Thereafter, you can obtain the EJBObject from this Handle.

Connector Known Issues

Change Request Number

Description

CR091818

Some customers would like to disable the proxy test that is automatically invoked the first time a connection to the back-end system is made.

This problem was resolved in 8.1 SP2.

EJB Known Issues

Change Request Number

Description

CR105964

When XML Spy is installed, DTD file types are automatically associated with XML Spy. Therefore, when customers open the DTD in a browser, XML Spy attempts to validate the DTD, and produces the following error:

=== Error message === The XML page cannot be displayed Cannot view XML input using style sheet. Please correct the error and then click the Refresh button, or try again later. ------------------------------------------------------------------------------- Cannot have a DTD declaration outside of a DTD. Error processing resource 'http://www.bea.com/servers/wls810/dtd/weblogic-rdbms20-persistence-810.dtd'. Line 13, Position 11 <!ELEMENT weblogic-rdbms-jar ( ----------^

Workarounds:

  • View the file in XML Spy (using the Tools —> Edit option). When viewing the file in XML Spy, it can also be saved on disk using Save As option.

  • Delete the DTD file type association (using the File Explorer —> Tools —> Folder Options —> File Types option).

CR106907

Some XA JDBC drivers do not support local transaction operations, which can cause an error similar to the following when optimistic concurrency is used with such a driver:

SQL operations are not allowed with no global transaction by default for XA drivers.

In other words, the error will occur when SupportsLocalTransactions="true" is specified for the JDBCConnectionPool.

This problem occurs because optimistic concurrency suspends a global transaction and does reads in a local transaction when the database is not Oracle. (When using Oracle, you can avoid this problem by explicitly setting <database-type>Oracle</database-type> in your CMP deployment descriptor.)

Workaround: Use the "RollbackLocalTxUponConnClose" on the JDBCConnection.

Examples Known Issues

Change Request Number

Description

CR072190

A known Tuxedo issue (CR050182) prevents Tuxedo clients and servers from compiling on HP platforms for the following example packages:

  • iiop.ejb.stateless.tuxclient

  • iiop.ejb.stateless.txtuxclient

  • iiop.ejb.entity.tuxclient

  • iiop.ejb.statelss.tux

Workaround: Use a platform other than HP.

CR108402

In Step 3 of the instructions in the package-summary.html file for the examples.jsp.tagext.form_validation example, the directory to which you must change is incorrect. The correct directory is SAMPLES_HOME\server\examples\src\examples\jsp\tagext\form_validation.

This problem was resolved in 8.1 SP2.

CR121588

The MedRec server fails to start on WebLogic Server 8.1 SP1 on Solaris 8 and Windows 2003 SP3 while attempting to connect to jarkarta.apache.org:80. This issue occurs because of a bug in Struts, documented at http://www.mail-archive.com/struts-user@jakarta.apache.org/msg60534.html.

Workaround: A connection to apache.org should be available during server startup.

JTA Known Issues

Change Request Number

Description

CR259037

When a transaction times out at logging time, you could experience a deadlock situation. The thread that is trying to write the transaction to the transaction log (TLOG) times out, a different thread trying to execute the rollback to clean the same transaction from the TLOG and deadlock each other.

Workaround or Solution:

Decrease the value of the CheckpointIntervalSeconds attribute and move TLOGs to a directory where they will not run out of memory.

JVM Known Issues

Change Request Number

Description

CR099962

When running on the IA64 platform with the JRockit JVM, non-trivial Web applications (that is, those with complex JSPs) experience performance problems when first loaded.

This problem occurs because JRockit is a compile-only JVM, and consequently, it has problems at first with very large methods such as those being generated from a complex JSP. The first time such a JSP is hit, the JVM will generate the native code for the method, and this can take ten of seconds. Note that reloads will not be slow. This problem typically occurs the first time a server is booted.

Workaround: Non-trivial Web applications either should not be run on JRockit on the IA64 platform, or they will just perform slower than usual (only on the first load).

Miscellaneous Known Issues

Change Request Number

Description

CR104392

Under stress conditions on Linux 64, sockets can be left in a CLOSE_WAIT state even after they are explicitly closed. This could possibly be due to socket buffer overflow, which may cause packet loss during the final TCP handshake to close the connection.

Workaround:

The workaround is to increase the amount of memory the system allocates for each TCP socket connection as follows:

  • tcp_rmem—128388607

  • tcp_wmem—128388607

  • rmem_max—128388607

  • wmem_max—128388607

  • tcp_keepalive_time—1800

  • tcp_window_scaling—0

  • tcp_sack—0

  • tcp_timestamps—0

Node Manager Known Issues

Change Request Number

Description

CR109093

When restarted, Node Manager requires the generated NodeManagerLogs/NodeManagerInternal/NodeManagerProperties file. It uses the seed in this file to decrypt the password.

Please do not delete this file. Users could accidentally delete this file since it is under the logs directory.

Plug-Ins Known Issues

Change Request Number

Description

CR102290

This version of WebLogic Server does not support Apache plug-ins for the Windows 64 platform because the Apache binary distribution is not yet available.

Security Known Issues

For information about BEA security advisories, refer to the BEA Advisories & Notifications page on the dev2dev Web site. On this page, you can download security-related patches and register to receive notifications of newly available security advisories. You can access the Advisories & Notifications page with the following URL: http://dev2dev.bea.com/advisories.

BEA has established an e-mail address (security-report@bea.com) to which you can send reports of any possible security issues in BEA products.

The following table summarizes security-related known issues for the 8.1 SP1 release.

Change Request Number

Description

CR100789

Using nCipher with JRockit for the IIOP thin client is not working properly.

Workaround: Do not use nCipher in this environment.

CR103330

The createPolicy method inherited from the PolicyEditorMBean optional SSPI is not case sensitive. Therefore, two resources with resourceIDs differing only by case can not be created within a WebLogic Authorization provider.

CR110389

The weblogic.Admin and weblogic.Deployer methods can be used with a username and password. If you specify a username on the command line but do not specify a password, they will prompt for one. By default, the password is not echoed to the screen as you type it. If you run the admin command without having set the client's path to include ...\weblogic81\server\bin in your PATH, then the native libraries that perform the non-echoing input cannot be found and weblogic.Admin and weblogic.Deployer will echo the password as you type it. By default, all WebLogic Serverstart scripts include this directory in the path so this is only an issue if you are using custom scripts or no scripts at all.

CR113469

Due to an unintentional change in the source code, security providers compiled for releases prior to WebLogic Server 8.1 will need to be recompiled for 8.1 to run.

Servlets & JSPs Known Issues

Change Request Number

Description

CR106460

Previous versions of WebLogic Server resolved URIs that contained extra spaces. WebLogic Server 8.1 does not resolve extra spaces, and a URI request that contains extra spaces will result in a 404. For example, previous versions of WebLogic Server resolved http://server:port/mywebapp/foo%20%20 to the resource foo in the Web application mywebapp, but beginning with 8.1 it no longer does.

CR111102

The save-sessions-enabled option to maintain HttpSession state across redeployments does not work if the session persistence type is set to Replicated.

CR262289

When calling the close() method on the InputStream obtained from the HttpsURLConnection class, the underlying SSL socket might not be closed even when the disconnect() method is called. This can result in a socket leak.

Workaround or Solution:

To avoid such a socket leak and ensure all underlying resources are closed, you are advised to use the HttpsURLConnection#disconnect() method instead of the InputStream#close() method.

System Administration Known Issues

Change Request Number

Description

CR110835

Configuration objects created during server startup are not always persisted.

If you use a startup class to create configuration objects (such as JDBC Connection Pools or any other resource whose configuration is represented by MBeans), WebLogic Server does not save the configuration data to config.xml unless you modify the configuration after the server startup cycle.

Workaround: Instead of using startup classes to create configuration objects, use weblogic.Admin, wlconfig, or the Administration Console after the server has completed its startup cycle.

WebLogic JDBC Type 4 Driver for MS SQL Server Known Issues

Change Request Number

Description

n/a

Before you can use distributed transactions with the WebLogic JDBC Type 4 Driver for MS SQL Server, you must install JDBC XA stored procedures in the database. See "Installing Stored Procedures for JTA" in BEA Weblogic Type 4 JDBC Drivers.

n/a

Microsoft SQL Server 7 does not allow resource sharing because it cannot release the connection to a transaction until it commits or rolls back. For example, the following code will fail:

xaResource.start(xid1, TMNOFLAGS)
...
xaResource.end(xid1, TMSUCCESS)
xaResource.start(xid2, TMNOFLAGS) ---> fail

Workaround: Set keepXAConnTillTxComplete=true on the JDBC connection pool.

n/a

When a global transaction is suspended or ended on a connection to a Microsoft SQL Server database, the database does not disassociate the transaction with subsequent database operations. If the global transaction is rolled back, all other database operations that were done since the transaction started will be rolled back with the transaction. For example, in the following code, the insert operation on Table2 is rolled back even though it is outside the scope of the transaction:

xaResource.start(xid1, TMNOFLAGS)
stmt.executeUpdate("insert into table1 values (1)");
xaResource.end(xid1, TMSUCCESS)

stmt.executeUpdate("insert into table2 values (2)");

xaResource.prepare(xid1);
xaResource.rollback(xid1);

n/a

When requesting an updatable ResultSet (by specifying ResultSet.CONCUR_UPDATEABLE) and a server-side cursor is created, the updatable ResultSet has pessimistic-locking semantics. If a server-side cursor is not created, the updatable ResultSet has optimistic-locking semantics. A server-side cursor may be created if selectMethod is set to cursor or if a scrollable ResultSet is requested. Ultimately, the transaction isolation level should be used to manage concurrency issues.

n/a

The WebLogic JDBC Type 4 Driver for MS SQL Server supports retrieval of output parameters from a stored procedure before all ResultSets or update counts have been completely processed. When CallableStatement.getXXX is called, ResultSets and update counts yet to be processed by the application are discarded to make the output parameter data available. Warnings are generated when results are discarded.

n/a

When attempting to create an updatable, scroll-sensitive ResultSet for a query that contains an expression as one of the columns, the driver cannot satisfy the scroll-sensitive request. The driver downgrades the type of the result returned to scroll-insensitive.

n/a

The preferred method for executing a stored procedure that generates ResultSets and update counts is using CallableStatement.execute(). If multiple results are generated using executeUpdate or executeQuery, the first update count or ResultSet, respectively, in the multiple result is returned. All other results are discarded. Warnings are generated when results are discarded.

n/a

The WebLogic JDBC Type 4 Driver for MS SQL Server allows PreparedStatement.setXXX methods and ResultSet.getXXX methods on BLOB/CLOB types above what is described in the JDBC specification. The supported conversions are generally the same as those for LONGVARBINARY/LONGVARCHAR, except where limited by database support.

n/a

If you are using Microsoft SQL Server 7, you should install Service Pack 2 or higher on the database server to ensure correct handling of character parameters.

n/a

Because of the way CHAR, VARCHAR, and LONGVARCHAR data types are handled internally by the driver, parameters of these data types exceeding 4000 characters in length cannot be compared or sorted, except when using the IS NULL or LIKE operators.

n/a

The ResultSet methods getTimestamp, getDate, and getTime return References to mutable objects. If the object reference returned from one of these methods is modified, then refetching the column using the same method will return the modified value. The value is only modified in memory; the database value is not modified.

WebLogic Workshop Known Issues

Change Request Number

Description

n/a

Although the WebLogic Workshop software is distributed with WebLogic Server software, neither the WebLogic Workshop IDE nor the WebLogic Workshop runtime framework is supported on IA64 systems on any operating system.

Web Services Known Issues

Change Request Number

Description

n/a

By default, the result of running clientgen in WebLogic Server 8.1 is .class files rather than .java files. This is a change in behavior from running clientgen in WebLogic Server 7.0. To make clientgen generate .java files in WebLogic Server 8.1, use the keepgenerated="true" attribute in clientgen. Documentation for this attribute is available at clientgen in Programming WebLogic Web Services.

CR100413

When using Web Services and two-way SSL, identity assertion is required whether or not the Web Service resource is protected. It should be possible to configure WebLogic Server to establish link-level trust without requiring identity assertion for Web Services that do not require authentication.

Workaround: Clients should only use certificates when required to access the service (to avoid unnecessary overhead) and care should be taken that all valid client certificates map to valid WebLogic Server users.

CR105388

When XML encodings are sent as a SOAP attachment, the message is not sent correctly.

CR107934

Even if an encryptKeyName is not provided as part of the <servicegen> task, an EncryptBody attribute with the value equal to "true" is added to the generated web-services.xml file. This indicates that the SOAP message should be encrypted.

Workaround: Edit the generated web-services.xml deployment descriptor for the Web Service to set the EncryptBody="false".

This problem was resolved in 8.1 SP3.

CR108646

The <servicegen> task does not support the "mergewithexistingws" attribute as described in the documentation.

Workaround: Specify different destEar and manually merge the two web-services.xml files.

WLEC Known Issues

Change Request Number

Description

CR106393

In 8.1, the classes packaged in wleorb.jar will conflict with those in the JDK (rt.jar). Customers who use WLEC (which is deprecated) will experience an exception when WLEC connections are attempted.

Workaround:

Prepend wleorb.jar to the JVM bootclasspath using -Xbootclasspath/p:$WL_HOME/lib/wleorb.jar to get WLEC working properly. Note, however, that doing this will cause problems for customers also wanting to use IIOP. In this case, customers should use WebLogic Tuxedo Connector instead.

WTC Known Issues

Change Request Number

Description

CR106305

CR111102

The examples/wtc/atmi/simpview example fails to run.

Workaround:

In the WebLogic Server Configuration section, add the infoenc VIEW table to the server classpath before completing Step 1.

One method is to update the server class path by modifying the startExamplesServer script located at SAMPLES_HOME\domains\examples. NT/2000 users modify the startExamplesServer.cmd script. Unix users modify the startExamplesServer.sh script.

Example: set CLASSPATH=%WEBLOGIC_CLASSPATH%;%CLASSPATH%;%SAMPLES_HOME%\server\examples\src\examples\wtc\atmi\simpview\infoenc

CR109849

WTC 8.1 does not handle VIEW buffers from Tuxedo 6.5 services. This problem occurs because of a logic error in one of the internal WTC routines that deal with Tuxedo 6.5 compatibility.

This problem was resolved in 8.1 SP2.

 


Known Issues in WebLogic Server 8.1

The following sections describe known issues in WebLogic Server 8.1:

Administration Console Known Issues

Change Request Number

Description

CR079374

If you add an application module such as a Web Application to a Enterprise Application (EAR) already deployed in your domain, you must delete the Enterprise Application from the domain configuration and then add it again in order for the module to become active.

CR080476

You designate a default Web Application with the Web Application's deployment descriptor. You no longer use the Administration Console. See Deploying a Default Web Application in the Administration Console Online Help.

CR088462

There are problems setting targets for EJB deployments using Internet Explorer version 5.0. Internet Explorer version 5.0 is not supported for the Administration Console. Please use version 5.5 or 6.0.

CR091141

Netscape version 4.79 does not work correctly on Linux AS 2.1. Netscape 4.79 on Linux AS 2.1 is not supported for use with the Administration Console.

Workaround: Use the Mozilla 1.0 browser, which is supported for Linux AS2.1.

CR099866

Administration Console extensions (see Extending the Administration Console) do not function correctly if you use enable a domain-wide administration port (see Enabling the Domain-Wide Administration Port in the Administration Console Online Help).

CR100159

The instructions for unlocking a user account in the Security section of the Administration Console online help are incorrect.

Workaround:

To unlock a user account:

    1. Expand the Servers-->Monitoring tab.

    2. Click the Security tab.

    3. In the Unlock User attribute, enter the user name for a user of this server who has been locked out.

    4. Click Apply.

    5. If the unlock was successful, a confirmation message appears at the top of the Monitoring-->Security page.

CR100460

When you use the Administration Console to clone a server that is a target of an existing JMS distributed queue or topic, the following exception is displayed in the console:

javax.management.InvalidAttributeValueException:
Illegal target

Although the newly cloned server is created, it does not appear in the Navigation Tree, but does appear in the summary server view.

Workaround: To successfully clone a server that is the target of an existing queue or topic, do not use the clone function. Instead, create a new server and manually configure the server.

CR101317

When using the Administration Console with Internet Explorer 5.5, you may see the following JavaScript error message: "Object does not support this property or method". Upgrading to Internet Explorer 6.0 corrects this problem. This error appears only occasionally on some computers where unspecified Internet Explorer patches have been applied.

Classloader Known Issues

Change Request Number

Description

CR094307

A ClassCastException might occur if you change the classloader hierarchy. For example, consider an application that has a Web application and an EJB. The Web application has a ejb-local-ref to the EJB. If the default classloader structure is used, the Web Application is able to access the EJB using the ejb-ref. However, if the classloader structure is changed so that the Web application classloader and the EJB classloader become peers, a ClassCastException is thrown, since the EJB class is no longer loaded from the common parent classloader.

Workaround: The two Web application and EJB must share classloaders or the called module must have a classloader that is an ancestor of the calling module. The server fails the deployment if you provide a local reference to a module in a classloader that is a peer or offspring of the calling module.

Configuration Wizard Known Issues

Change Request Number

Description

n/a

When configuring a JDBC connection pool, you can only test the connection configuration parameters if you are using a JDBC driver installed with WebLogic Server.

Workaround:

To test connection configuration parameters for other drivers, you can change the script that launches the Configuration Wizard to include other drivers in the CLASSPATH. Follow these steps:

    1. Open WL_HOME\common\bin\config.cmd in a text editor (where WL_HOME is the folder where you installed WebLogic Server, typically C:\bea\weblogic81).

    2. Find the following lines:

if "%1" NEQ "" goto java

"%JAVA_HOME%\bin\javaw" %JAVA_VM% -jar "c:\bea\weblogic81\common\lib\config.jar"

goto end

:java

"%JAVA_HOME%\bin\java" %JAVA_VM% -jar "c:\bea\weblogic81\common\lib\config.jar" %*

    3. Replace those lines with the following single line:

"%JAVA_HOME%\bin\java" %JAVA_VM% -cp .;config.jar;3rdparty.jar;comdev.jar;wizard.ja r;%WL_HOME%\server\lib\weblogic.jar;%WL_HOME%\ common\eval\pointbase\lib\pbserver44.jar;%WL_H OME%\common\eval\pointbase\lib\pbtools44.jar;% WL_HOME%\common\eval\pointbase\lib\pbclient44. jar com.bea.plateng.wizard.WizardController

    4. Add the path to additional Type 4 JDBC drivers to the end of the -cp string (after pbclient44.jar).

Note: This change only affects the Configuration Wizard when you start it using the config.cmd file. It does not take affect when you start the Configuration Wizard from the Start menu.

These instructions are for a Windows platform. You can adapt them for a UNIX platform.

CORE Known Issues

Change Request Number

Description

CR101946

If you use the Java socket reader implementation, extended use of JMS can lead to an Out of Memory error.

Workaround: Use the native WebLogic Server performance pack. See WebLogic Server Performance and Tuning.

CR102874

By default, network channel tunneling is on, which may pose a security risk if you have explicitly configured network channels for a WebLogic Server instance.

This problem is resolved in 8.1 SP1, where network channel tunneling is disabled by default.

Deployment Known Issues

Change Request Number

Description

CR096834

If a member of a WebLogic Server cluster shuts down or becomes unavailable when you are deploying a large application to the cluster, the deployment task can appear to hang. This occurs because the two-phase deployment protocol is waiting for a response from the server that has shut down. The task will appear to hang until the default timeout length of 1 hour is reached.

Workaround: If you anticipate changes to a cluster during a deployment, use the -timeout option to weblogic.Deployer to specify a shorter timeout period.

CR100540

WebLogic Server uses inconsistent module names when you deploy archive files and exploded archive directories. When you deploy an archive file, its module names are not registered using the file extension. For example, consider the following case where you deploy an Enterprise Application, myapp.ear, that contains a module, myejb.jar:

java weblogic.Deployer -adminurl http://localhost:7001 -username weblogic -password weblogic -name myappdeployment -source ./myapp.ear -deploy

If you later try to target the EJB module to a new server using its package name, you will receive an error:

java weblogic.Deployer -adminurl http://localhost:7001 -username weblogic -password weblogic -name myappdeployment -targets myejb.jar@newserver -deploy

Application, myappdeployment, does not define the module, myejb.jar. Respecify the source to deploy additional modules for an application.

This occurs because the myejb.jar module is named myejb, rather than myejb.jar.

Workaround: Deploy using exploded archive directories.

EJB Known Issues

Change
Request
Number

Description

CR061938

In certain conditions, EJB QL queries can return spurious duplicates that are the results of SQL cross products. This can occur under the following conditions:

  • The EJB-QL query contains path expressions that navigate relationships; this generates multiple tables in the generated SQL SELECT clause.

  • The WHERE clause contains OR operands which navigate relationships, and not all of the path expression in the OR operands map to all of the tables in the generated SQL SELECT clause for the query; this may cause a cross product to show up in the results for that OR operand.

The following example illustrates the problem:

EJB QL:

SELECT OBJECT(c)
FROM CustomerBean AS c, IN(c.accounts)accts
WHERE c.name = '100' OR c.accts.bal = 300

DATA: customer '100' exists but has no accounts

EXPECTED RESULT: customer '100' from clause #1

ACTUAL RESULT: customer '100' X number of accts

SQL:

DROP TABLE thorick_customers;

CREATE TABLE thorick_customers (
cust_name VARCHAR(10),
cust_interests VARCHAR(10),
cust_rating INTEGER, acct_id INTEGER,
PRIMARY KEY (cust_name));


DROP TABLE thorick_accounts;

CREATE TABLE thorick_accounts
(acct_id INTEGER,
bal FLOAT,
PRIMARY KEY (acct_id));


INSERT INTO thorick_accounts VALUES (100, 100.0);
INSERT INTO thorick_accounts VALUES (200, 200.0);
INSERT INTO thorick_accounts VALUES (300, 300.0);
INSERT INTO thorick_accounts VALUES (400, 400.0);
INSERT INTO thorick_accounts VALUES (500, 500.0);
INSERT INTO thorick_customers VALUES('100', 'jazz', 2, null);
INSERT INTO thorick_customers VALUES('900', 'punk', 3, 400);


SELECT WL0.cust_name, WL0.cust_interests, WL0.acct_id
FROM thorick_Customers WL0, thorick_Accounts WL1
WHERE WL0.cust_name = '100'
OR ( wl1.bal = 300 AND wl0.acct_id=wl1.acct_id );


CUST_NAME CUST_INTER ACCT_ID
---------- ---------- ----------
100 jazz
100 jazz
100 jazz
100 jazz
100 jazz

CR093615

There is a known issue with removal of stateful session beans in a cluster. Attempts to remove the bean with EJBHome.remove result in error messages such as this:

<Dec 19, 2002 4:51:46 PM PST> <Info> <EJB> <BEA-010049> <EJB Exception in method : remove: java.rmi.NoSuchObjectException: Bean has been deleted.

Workaround: Use EJBObject.remove instead of EJBHome.remove to remove a stateful session bean that is deployed in a cluster.


 

Examples Known Issues

Change Request Number

Description

CR068247

The JSPs included with WebLogic Server were not being pre-compiled before they are accessed for the first time. If you were deploying precompiled JSPs, WebLogic Server may recompile them to the /wlnotdelete/ directory.

Workarounds:

    1. Precompile classes into you WEB-INF/classes directory (or a jar file in WEB-INF/lib) instead.

    2. Set a workingDir for the jsp-descriptor in your weblogic.xml:

<jsp-descriptor>
   <jsp-param>
     <param-name>workingDir</param-name>
     <param-value>d:\jsp_store</param-value>
   </jsp-param>
</jsp-descriptor>

CR099174

When you configure new domains and servers in a WebLogic Server 8.1 installation, the MedRec Samples Authenticator is available as a provider for the security realm. However, this provider is only usable in a domain that includes the MedRec sample application and PointBase database.

If you want to remove the MedRec Samples Authenticator from a newly-installed domain:

1) Delete the wlMedRecSampleAuthProvider.jar file from the WL_HOME\server\lib\mbeantypes directory.

2) In the Administration Console, go to the Security->Realms->Providers->Authentication node and delete the MedRec Samples Authenticator reference.

Note: If you follow these steps to remove the provider, you will not be able to log into the Patient Web Application of the MedRec sample application. The Patient application requires the MedRec Sample Authenticator to query the MedRec RDBMS for patient usernames, passwords, and assigned groups.

CR101326

The C++ clients for IIOP examples are not provided in this release. You will not be able to build or reference documentation for these clients until they are provided in a later release of the product, or on the http://dev2dev.bea.com site.

This problem was resolved in 8.1 SP1.

CR101713

When running examples.cluster.ejb example, failover did not occur properly. Instead, the secondary server threw this exception:

Trouble while getting database connection java.sql.SQLException: Pool connect failed: java.lang.Exception: Weblogic Pool Driver doesn't support XA driver, Please change your config to use a Non-XA driver at weblogic.jdbc.common.internal.JDBCUtil.wrapAndThrowResourceException(JDBCUtil.java:161) at weblogic.jdbc.pool.Driver.connect(Driver.java:155 at weblogic.jdbc.jts.Driver.getNonTxConnection(Driver.java:394) at weblogic.jdbc.jts.Driver.connect(Driver.java:137) at .....

The error occurs because of an error in the sample instructions. package.html contained incorrect instructions for configuring the connection pool. The instructions say to use the default XA Pointbase pool. Instead, the pool should be configured as a non-XA pool. These instructions are corrected in WebLogic Server 8.1 SP2.

Installation Known Issues

Change Request Number

Description

CR101518

For the Japanese version of the installation, the shortcut to Online Documentation from the Windows Start menu linked to the English version of e-docs, rather than to the Japanese version of e-docs.

Interoperability Known Issues

Change Request Number

Description

CR092181

Failover for entity beans and EJB handles requires that the cluster address be specified as a DNS name that maps to all server instances in the cluster and only server instances in the cluster. The cluster DNS name should not map to a server instance that is not a member of the cluster.

CR095479

Remote methods throwing MarshalException/UnmarshalException when interoperating between 7.0 SP1 and 8.1 servers may result in unmarshalling problems. This occurs because MarshalException and UnmarshalException map to the correct completion statuses in 8.1 but the incorrect completion statuses in 7.0 SP1.

CR100713

WebLogic Server 6.1 SP04 and earlier does not interoperate transactionally in some scenarios with WebLogic Server 8.1 over t3. Transactions between 6.1 and 8.1 server instances over t3 timeout with an ArrayIndexOutOfBoundsException on the 6.1 server instance.

CR101186

Exceptions thrown by WebLogic Server 8.1 cannot be decoded by WebLogic Server 7.0 SP02 or earlier. WebLogic Server 7.0 will throw a MARSHAL exception.

CR101324

Binding objects built with WebLogic Server 6.1 classes and deploying them to WebLogic Server 7.0 or higher is not supported and can result in failures during startup.

WebLogic Server 8.1 Interoperability Notes

WebLogic Server 8.1 does not support mixed-version domains. All Managed Servers in a domain must be at the same version (8.1).

WebLogic Server 6.1 has some limitations when acting as a client to WebLogic Server 8.1 over the IIOP protocol. This is because WebLogic Server 6.1 does not support class evolution, so only identical classes can be passed between WebLogic Server 6.1 and WebLogic Server 8.1. These issues were partially resolved through the support of class evolution in WebLogic Server 7.0 and fully resolved in WebLogic Server 7.0 SP3. For WebLogic Server 7.0 IIOP clients interoperating with WebLogic Server 8.1, customers should use the WLS RMI-IIOP runtime by specifying -Dweblogic.system.iiop.enableClient="true" on the client command-line. This option is not supported in WebLogicServer 6.1 and not necessary in WebLogic Server 8.1.

In order to interoperate with WebLogic Server version 6.1, you must add the following elements to the <Domain> section of the WebLogic Server 8.1 config.xml file:

<!-- START OF 70 INTEROP SPECIFIC SETTING -->
<Security CompatibilityMode="true" GuestDisabled="false"
InteropEnabled="true"
InteropPassword="{3DES}OuQN48TIXRYFVVez4VBFTA=="
InteropUsername="system" Name="@DOMAIN1"
PasswordPolicy="wl_default_password_policy" Realm="wl_default_realm"
RealmSetup="true" SystemUser="system"/>
<!-- START PER CR071185, ADDING BELOW SETTING TO 70 ONLY -->
<SecurityConfiguration Credential="gumby1234" Name="@DOMAIN1" />
<!-- END PER CR071185, ADDING BELOW SETTING TO 70 ONLY -->
<!-- END OF 70 INTEROP SPECIFIC SETTING -->

(The above elements are also required for WebLogic Server version 7.0 to interoperate with WebLogic Server version 6.1.)

JVM Interoperability

When WebLogic Server instances interoperate with other WebLogic Server instances, each server instance must use a certified JVM. For example, to interoperate with WebLogic Server 8.1, WebLogic Server versions 6.1 and 7.0 are certified to use Sun's JDK version 1.3.1_x, while WebLogic Server 8.1 is certified to use JDK version 1.4.1_02. See Supported Configurations for more information on certified JVMs.

For standalone clients (rather than server instances) the same JVM considerations apply.

WebLogic Server 8.1 running on JDK 1.4.1 will interoperate with 7.0 SP2 running on JDK 1.3.1.

J2EE Known Issues

Change Request Number

Description

CR076513

You cannot deploy a standalone Enterprise Application module (Web Application, EJB, or Resource Adapter) when an unrelated META-INF/application.xml file resides in the same directory. For example, you cannot deploy a .rar file from c:\mydeployment\myadapter.rar if an unrelated c:\mydeployment\META-INF\application.xml file exists.

Workaround: Delete the unrelated META-INF directory or place the module in a dedicated directory.

JDBC Known Issues

Change Request Number

Description

CR177621

PointBase Server is an all-Java DBMS product included in the WebLogic Server distribution solely in support of WebLogic Server evaluation, either in the form of custom trial applications or through packaged sample applications provided with WebLogic Server. Non-evaluation development and/or production use of the PointBase Server requires a separate license to be obtained by the end user directly from PointBase.

n/a

The connection creation retries option for a connection pool depends on the message returned from the JDBC driver to determine whether to continue to try to create database connections. In some cases, Weblogic Server will not retry to create database connections.

Workaround: Correct the problem with your database, and then untarget and retarget the connection pool in the Administration Console. See Deploying a JDBC Connection Pool to One or More Servers or Clusters in the Administration Console Online Help.

For more information about the connection creation retries option, see the Connection Creation Retry Frequency page in the Administration Console Online Help and Enabling Connection Creation Retries in Programming WebLogic JDBC.

n/a

The JDBC Connection Pool Assistant in the Administration Console automatically sets the Test Table Name attribute for a connection pool based on the DBMS of the JDBC driver that you select. The Test Table Name attribute is used in connection testing which is optionally performed periodically or when you create, reserve, or release a connection, depending on how you configure the connection pool. For the test to succeed, the database user used to create database connections in the connection pool must have access to the database table. If not, you should either grant access to the user (make this change in the DBMS) or change the Test Table Name attribute to the name of a table to which the user does have access (make this change in the WebLogic Server Administration Console). For a list of default values, see "Default Test Table Name" in the Administration Console Online Help.

CR082336

In WebLogic Server 8.1, dbKona was removed from the distribution. It was deprecated in previous releases.

CR086874

Processes can no longer share a reference to a connection from a connection pool. For example, a JSP cannot request and receive a connection from an EJB. This was allowed in previous releases of WebLogic Server, but is no longer allowed because it occasionally resulted in multiple threads accessing a single connection or invalid references because a connection was closed or garbage collected.

CR086902

For connection pools that connect to a PointBase database, statement caching is disabled by default because PointBase does not effectively handle cached prepared or callable statements after a table has been dropped and recreated. If your code does not include such DDL operations, you can enable statement caching in the Administration Console. See Increasing Performance with the Statement Cache page in the Administration Console Online Help.

CR093175

The JDBCDataSourceMBean is deprecated in WebLogic Server 8.1. Use the JDBCTxDataSourceMBean instead. The attributes that are not available in the JDBCTxDataSourceMBean (WaitForConnectionEnabled and ConnectionWaitPeriod) have been deprecated and are replaced with the ConnectionReserveTimeoutSeconds attribute in the JDBCConnectionPoolMBean.

CR099442

The documentation regarding Emulate Two Phase Commit in the Administration Console Online Help as installed with the WebLogic Server software is incorrect. This feature does not use the last-agent commit optimization, as documented. The correct documentation is available in the e-docs version of the Administration Console Online Help.

CR100625

Several weblogic.Admin JDBC commands behave differently than documented in Commands for Managing JDBC Connection Pools in the WebLogic Server Command Reference. The documentation describes the desired behavior, rather than the actual behavior. The commands will be fixed in a future release to match the documentation. Commands with behavior other than what is documented:

  • SUSPEND_POOL - This command does not take true or false parameters. Instead, it defaults to the behavior for the false option. Syntax:

java [SSL trust options] weblogic.Admin
[ [-url | -adminurl]
URL] -username username -password password SUSPEND_POOL -poolName connection_pool_name

  • SHUTDOWN_POOL - This command does not take true or false parameters. Instead, it defaults to the behavior for the false option. Syntax:

java [SSL trust options] weblogic.Admin
[ [-url | -adminurl]
URL] -username username
-password
password SHUTDOWN_POOL -poolName
connection_pool_name

CR101419

Fail-over for DataSources and TxDataSources in a cluster is not working in this release. This problem is resolved in Service Pack 1.

JTA Known Issues

Change Request Number

Description

CR259037

When a transaction times out at logging time, you could experience a deadlock situation. The thread that is trying to write the transaction to the transaction log (TLOG) times out, a different thread trying to execute the rollback to clean the same transaction from the TLOG and deadlock each other.

Workaround or Solution:

Decrease the value of the CheckpointIntervalSeconds attribute and move TLOGs to a directory where they will not run out of memory.

JMS Known Issues

Change Request Number

Description

CR097038

When a source JMS bridge destination is not running in the same WebLogic Server 8.1 domain as the Message Bridge, then a trusted security relationship must be established between the domains. Otherwise, the Messaging Bridge must be configured to work in synchronous mode (Asynchronous Mode Enabled = No).

For more information on configuring security interoperability for the Messaging Bridge, see "Messaging Bridge" in the Administration Console Online Help.

CR098975

WebLogic Server sometimes fails to boot after you undeploy a distributed destination. This occurs when you delete the distributed members and the distributed destination itself, but do not delete the underlying physical destinations. The following sequence of Administration Console actions will cause this problem:

    1. In the Configuration -> Members tab of a distributed destination, click the trash can icon to remove a member.

    2. When prompted, do not delete the member's underlying physical destination.

    3. Click Remove to delete the member.

    4. If necessary, continue deleting all the members of the distributed destination.

    5. Delete the distributed destination.

If you shut down the server at this point, you will not be able to reboot it.

Workaround: Delete each distributed destination member's underlying physical destination when prompted by the Administration Console.

JVM Known Issues

Change Request Number

Description

CR099314

There is a problem with JDK 1.3.1_04 and the nohup command, as described at http://developer.java.sun.com/developer/bugParade/bugs/4755829.html. The nohup command starts failing; the WebLogic Server process exits when the parent shell exits, regardless of the invocation of nohup. The problem was confirmed for ksh (the default Solaris 8 shell), but may exist for other shells. The problem does not occur with the bash shell.

Workarounds:

  • Use "(startWeblogic >output 2>&1 &)" instead of "nohup startWeblogic &".

Oracle Thin Driver Known Issues

Note: In addition to the issues listed below, see Known Oracle Thin Driver Issues in Programming WebLogic JTA. That section contains information about known problems in the Oracle Thin Driver 9.2.0, and ways to work around the problems. WebLogic Server 8.1 ships with the 9.2.0.2 version of the Oracle Thin Driver, which corrects some known issues not listed here or in Programming WebLogic JTA.

Change Request Number

Description

n/a

After transaction recovery has completed from a server restart, some customers have seen frequent occurrences of the ORA-1591 exception, which may be followed by ORA-24756 or ORA-24776 exceptions.

Workarounds:

If you see these errors, you should refer to the following Oracle notes: Note:100664.1 and Doc ID: 231196.1.

You should also ensure that the application is not leaking connections from the JDBC connection pool. You can check by:

    1. Enabling Connection Leak Profiling on the JDBC Connection Pool --> Configuration --> Connections tab in the Administration Console (under Advanced Options).

    2. In the navigation tree in the Administration Console, right-click the connection pool used in the application and select View Leaked Connections.

For more information, see Enable Connection Leak Profiling in the Administration Console Online Help.

If you continue to see ORA-1591 errors, you may want to contact Oracle Support for help with tuning your database. Before contacting Oracle Support, consider doing the following:

  • Generate a trace file for the ORA-1591.

  • Run the statspack report and analyze contentions in the database.

CR101709

Application code can inadvertently start local transactions on a Connection object, which can cause XAER_PROTO errors. According to the SQL spec, any DDL/DML operation can potentially start a local transaction. It is the application's responsibility to commit or rollback these transactions.

This was not a problem with earlier versions of the Oracle Thin driver because earlier versions of the driver did not check for active local transactions before starting a global transaction or active global transactions before starting a local transaction. However, version 920 of the Oracle Thin driver has started checking for this. It throws an XAER_PROTO error if an application (or WebLogic Server) tries to start a global transaction (XAResource.start()) on a connection on which a local transaction is already active.

Workaround:

The application code must end the local transaction as an additional step before closing the connection (which returns the connection to the connection pool).

This problem was resolved in 8.1 SP1.


 

RMI/RMI-IIOP Known Issues

Change Request Number

Description

CR095810

You cannot use callbacks through a firewall when using the thin client .jar files (wlclient.jar and wljmsclient.jar) and IIOP or IIOPS.

Workaround: Callbacks do work when using tunneling.

This problem was resolved in 8.1 SP1.

CR101476

An applet using the thin client .jar files (wlclient.jar and wljmsclient.jar) cannot perform tunneling using the Sun Java2 SDK 1.4.1_02 installed with WebLogic Server. This problem is fixed in Sun's SDK versions 1.4.1_03 and 1.4.2.

CR124265

WebLogic Server 8.1 provides its own ORB implementation in place of the ORB shipped with J2SE 1.4. This ORB is instantiated by default when programs call ORB.init(), or when "java:comp/ORB" is looked up in JNDI. Customers who wish to use a different ORB by default should set the following properties:

org.omg.CORBA.ORBSingletonClass=<classname>
org.omg.CORBA.ORBClass=<classname>

Note that ORBSingletonClass must be set on the server command-line, whereas ORBClass can be set as a property argument to ORB.init(). Customers wishing to use a different RMI-IIOP implementation must set the following two properties:

javax.rmi.CORBA.UtilClass=<classname>
javax.rmi.CORBA.PortableRemoteObjectClass=<classname>

Customers who do this will get the following errors at server startup:

<Sep 19, 2003 9:12:03 AM CDT> <Error> <IIOP> <BEA-002015> <Using javax.rmi.CORBA.UtilClass <classname>; The IIOP subsystem requires a WebLogic Server-compatible UtilClass.>

<Sep 19, 2003 9:12:03 AM CDT> <Error> <IIOP> <BEA-002016> <Using javax.rmi.CORBA.PortableRemoteObjectClass <classname>, the IIOP subsystem requires a WebLogic Server-compatible PortableRemoteObjectClass.>

indicating that the WebLogic RMI-IIOP runtime will not work.

The J2SE defaults for these properties are:

org.omg.CORBA.ORBSingletonClass=com.sun.corba.se.internal.corba.ORBSingleton
org.omg.CORBA.ORBClass=com.sun.corba.se.internal.Interceptors.PIORB
javax.rmi.CORBA.UtilClass=com.sun.corba.se.internal.POA.ShutdownUtilDelegate
javax.rmi.CORBA.PortableRemoteObjectClass=com.sun.corba.se.internal.javax.rmi.PortableRemoteObject

This problem was resolved in 8.1 SP2.

CR186728

JMS IIOP thin client fails with the following exception:

JMSServer "add JMS Dispatcher peerGone". Unhandled exception.

For more information, see Setting Network Channel Addresses.

Security Known Issues

For information about BEA security advisories, refer to the BEA Advisories & Notifications page on the dev2dev Web site. On this page, you can download security-related patches and register to receive notifications of newly available security advisories. You can access the Advisories & Notifications page with the following URL: http://dev2dev.bea.com/advisories.

BEA has established an e-mail address (security-report@bea.com) to which you can send reports of any possible security issues in BEA products.

The following table summarizes security-related known issues for the 8.1 release.

Change Request Number

Description

CR090469

If you are upgrading from WebLogic Server 6.1 or earlier and you want to use the security features that were introduced in WebLogic Server 7.0, your domain's config.xml file must declare a configuration version of 7.0 or later. For example,

<Domain ConfigurationVersion="8.1.0.0" Name="mydomain">

If you are using a config.xml file that you created with WebLogic Server 6.1 or earlier, you must open config.xml in a text editor and add ConfigurationVersion="8.1.0.0" to the <Domain> element.

CR091648

CR099358

The MedRec Authentication provider implements authentication for the MedRec sample applications and cannot be used in a production domain.

Workaround:

Delete the JAR file for the MedRec Authentication provider (wlMedRecSampleProvider) from the WL_HOME\lib\mbeantypes directory and reboot WebLogic Server. The MedRec Authentication provider should no longer appear in the list of available Authentication providers on the Providers-->Authentication Providers tab in the WebLogic Server Administration Console.

CR097038

When using the Messaging Bridge, non-trusted domains will only work when using synchronous mode communications.

Workaround: WebLogic Server only supports interoperability between domains for synchronous operations.

CR100783

The following exception may occur if the password to unlock the private key was entered incorrectly:

ImportPrivateKey failed,
java.security.KeyManagementException:ASN.1:
Lengths longer than 32 bits are not supported
java.security.KeyManagementException:ASN.1:
Lengths longer than 32 bits are not supported at
com.certicom.security.cert.internal.x509.SSLPlusSupport.
getLocalIdentityPartial(Unknown Source) at com.certicom.net.ssl.CerticomContextWrapper.
inputPrivateKey(Unknown Source) at utils.ImportPrivateKey.importKey(ImportPrivateKey.java:67) at utils.ImportPrivateKey.main(ImportPrivateKey.java:23)

It is also possible to get this exception if the private key was corrupt or was some format that WebLogic Server does not recognize.

Workaround: Re-enter the correct password. If you have forgotten the password for the private key or if the private key is corrupt, obtain a new private key.

This problem was resolved in 8.1 SP1.

CR101274

Please review the security advisory information at http://www.ncipher.com/support/advisories.

This problem was resolved in 8.1 SP1.

CR101299

Figure 4-7, Auditing Process, in the Introduction to WebLogic Security document is incomplete in as much as it does not show all the Audit events generated by WebLogic Server.

Not only are Audit events written for authentication success and failure (when identity is authenticated, impersonated, and asserted), Audit events are also written for Authorization provider exceptions, Adjudication provider decisions, user lockout unlock, locked, and lockout expired actions, Role Mapping provider exceptions, and when security roles are obtained.

Figure 4-7 is correct but incomplete in that it only refers to authentication and does not imply that an authentication action can result in different audit events depending upon what happens.

CR101770

The SSL Login Timeout attribute was unintentionally dropped from the WebLogic Server Administration Console. Support for the attribute was added in WebLogic Server 8.1 SP1.

This problem was resolved in 8.1 SP1.

Servlets & JSPs Known Issues

Change Request Number

Description

CR100068

CR110324

You can not use JSTL tags in JSPs that include Japanese characters. When such a JSP is executed, an error starting with the following lines occurs:

java.io.IOException: javax.servlet.jsp.JspException: The taglib validator rejected the page: "org.xml.sax.SAXParseException: An invalid XML character (Unicode: 0x82) was found in the CDATA section."

...

This problem occurs because of the following, conflicting conditions:

    1. The page encoding of JSPs is defined by Shift_JIS, as in:
    <%@ page pageEncoding="Shift_JIS" %>

    2. Using multibyte characters (Japanese) in JSPs.

    3. Using JSTL tags.

CR262289

When calling the close() method on the InputStream obtained from the HttpsURLConnection class, the underlying SSL socket might not be closed even when the disconnect() method is called. This can result in a socket leak.

Workaround or Solution:

To avoid such a socket leak and ensure all underlying resources are closed, you are advised to use the HttpsURLConnection#disconnect() method instead of the InputStream#close() method.

System Administration Known Issues

Change Request Number

Description

CR085065

The WebLogicMBean.setName method, which is inherited by all WebLogic Server MBeans, is no longer supported. The method changes the Name component of an MBean instance's WebLogicObjectName. Because this component is part of the primary key for a WebLogicObjectName, changing the value leads to undesired behavior.

Instead of setting the value of an MBean's Name attribute, the API throws InvalidAttributeValueException.

Workaround: If you want to change the name of an MBean instance, destroy the current instance and create a new one with a new name.

CR094188

If you place an application in the servername/applications directory and then start the server in development mode, the server automatically deploys the application, but it does not save the application configuration to the config.xml file. If you start the server in production mode, the server neither deploys the application nor saves the configuration to the config.xml file.

Workaround: Start the server first and then deploy the application. In development mode, you can deploy the application by copying it to the servername/applications directory.

CR100089

Within a domain, each server, machine, cluster, JDBC connection pool, virtual host, and any other resource type must be named uniquely and must not use the same name as the domain.

If you create a resource with a non-unique name, WebLogic Server emits an error message.

This requirement applies to the Name component of WebLogicObjectName as well. For example, if you use WebLogic Server APIs to create Configuration MBeans, the value of the Name component must be unique amongst all other MBeans in the domain.

Tools Known Issues

Change Request Number

Description

CR097668

When using cygwin version 1.3.20(0.73/3/2) the shell scripts fail if the cygdrive prefix has been changed (for example, mount -c / so /c/windows instead of /cygdrive/c/windows). This is because the value '/cygdrive' is hardcoded in the scripts. The following scripts are affected:

weblogic700/common/bin/commEnv.sh

weblogic700/server/bin/setWLSEnv.sh

weblogic700/server/bin/startNodeManager.sh

weblogic700/server/bin/startWLS.sh

Workaround: Replace the value of /cygdrive in these scripts to get the correct cygdrive prefix:

CYGDRIVE=\Qmount -ps | tail -1 | awk '{print $1}' | sed -e 's%/$%%'\Q

For example, in startWLS.sh find:

    if [ -n "\Quname -s |grep -i cygwin_\Q" ]; then
    WL_HOME=\Qecho $WL_HOME | sed "s#\([a-zA-Z]\):#/cygdrive/\1#g"\Q
    JAVA_HOME=\Qecho $JAVA_HOME | sed "s#\([a-zA-Z]\):#/cygdrive/\1#g"\Q
    fi

and change it to:

    if [ -n "\Quname -s |grep -i cygwin_\Q" ]; then
    CYGDRIVE=\Qmount -ps | tail -1 | awk '{print $1}' | sed -e 's%/$%%'\Q
    WL_HOME=\Qecho $WL_HOME | sed "s#\([a-zA-Z]\):#${CYGDRIVE}/\1#g"\Q
    JAVA_HOME=\Qecho $JAVA_HOME | sed "s#\([a-zA-Z]\):#${CYGDRIVE}/\1#g"\Q
    fi

Change the other scripts in a similar manner.

WebLogic jDriver for Oracle Known Issues

Change Request Number

Description

n/a

You cannot use windows-31j encoding with the BEA WebLogic JDriverTM for Oracle because of a known limitation in the Sun Java 2 SDK 1.4.1 (Bug Id 4715330): the windows-31j converter is missing in lib/charsets.jar. The encoding of WebLogic jDriver for Oracle is based on lib/charsets.jar, so this character encoding is not supported on the Sun Java 2 SDK 1.4.1. Sun intends to fix this problem in the Java 2 SDK 1.4.1_05 and 1.4.2 release. BEA fixed this problem in the WebLogic JRockit 8.1 Service Pack 1.

WebLogic Tuxedo Connector Known Issues

Change Request Number

Description

CR067275

Load balancing and failover is not supported with WebLogic Tuxedo Connector in a cluster using RMI over IIOP.

CR079630

Most WebLogic Tuxedo Connector tBridge users will expect to set the Correlation ID using the setJMSCorrelationID(String) method. This will take a 32 character string and turn it into a 64 byte array. JMS stores the string as UTF-16BE. When the tBridge receives the Correlation Id from Tuxedo it is 32 characters as 32 bytes. The tBridge then uses the setJMSCorrelationIDAsBytes(byte[]) method to set the message for JMS receiving queue. The characters may be the same in ASCII but the two will not compare due the length difference.

Workaround:

When it is necessary to compare the Correlation ID with the ID returned from Tuxedo in the receiving JMS queue, create a byte array containing the hexadecimal values of the Correlation ID. Then use the setJMSCorrelationIDAsBytes() and getJMSCorrelationIDAsBytes()methods to create an ID to compare with the ID returned from Tuxedo in the receiving JMS queue.

For example, if the string Correlation ID is "1234567890ABCDEFGHIJKLMNOPQRSTUV":

private byte[] coridbyte={0x31,0x32,0x33,0x34,0x35,0x36,0x37,
                                          0x38,0x39,0x30,0x41,0x42,0x43,0x44,
                                          0x45,0x46,0x47,0x48,0x49,0x4a,0x4b,
                                          0x4c,0x4d,0x4e,0x4f,0x50,0x51,0x52,
                                          0x53,0x54,0x55,0x56};
msg.setJMSCorrelationIDAsBytes(coridbyte);
corIDasBytes = msg.getJMSCorrelationIDAsBytes();

corIDasBytes contains the correct value to compare with the Correlation ID returned from Tuxedo.

CR095050

When using the WTCMigrateCF utility to migrate WebLogic Server 6.1 WTC configuration files, Weblogic Server throws NameNotFoundException when using the -deploy option. Users should ignore this exception in the server log.

CR095807

Linux and AIX compilers may produce compile errors when using Tuxedo based IDL with some value types. This issue is resolved in Tuxedo 8.1 RP001.

WebLogic Workshop Known Issues

Change Request Number

Description

n/a

WebLogic Server 8.1 ships with a Beta version of the WebLogic Workshop IDE and development framework. In the Beta release, WebLogic Workshop requires 1GB of system RAM.

n/a

WebLogic Workshop is not available in the Japanese version of WebLogic Server 8.1.

CR097002

WebLogic Server 8.1 ships with two JSP compilers: the original WebLogic JSP compiler, and a compiler that is optimized for use with WebLogic Workshop. Internationalization for the WebLogic Workshop-optimized JSP compiler will be completed after the initial release of WebLogic Server 8.1.

CR102094

The script used to start the WebLogic Workshop sample server, WL_HOME\samples\workshop\startWebLogic.cmd is not installed in this release. You cannot use the Windows Start Menu selection to start the sample server because it refers to this missing file.

Web Services Known Issues

Change Request Number

Description

CR063817

A complex type which contains an element of type xsd:schema causes an exception during autotyping.

CR074740

WebLogic Server does not convert Java data structures which contain Lists as items of Lists to an XML Schema.

CR083527

The typeMappingBase attribute of autotyper is ignored when generating holder classes.

CR086326

The -Dweblogic.webservice.verbose=true does not work for SSL clients. The workaround to get verbose output is to use -Dweblogic.webservice.binding.verbose=true.

CR092184

A service configured to use the security extensions cannot be accessed from the Web service test page.

CR101160

Complex data types in service-specific exceptions are not handled in strict accordance with the Java API for XML-Based RPC (JAX-RPC) 1.0 specification, as follows:

  • Section 4.3.6 WSDL Fault, Service Specific Exception: When the clientgen Ant task generates the JAX-RPC stubs from a WSDL that contains a wsdl:fault element whose message part is an xsd:complexType, the generated stubs incorrectly map the xsd:complexType to a simple Java type. This means that when a client application uses the stubs to invoke the Web Service, and the service throws the service-specific exception (rather than the a javax.xml.rpc.SOAPFaultException), the client application cannot access the full complex data from the exception.

There is no workaround to this problem.

  • Section 5.5.5 Methods, Exceptions, fourth bullet: The WebLogic Web Service runtime incorrectly maps multiple fields in a service-specific Java exception class to a simple XML data type, rather than to an xsd:complexType.

Workaround: Always throw a javax.xml.rpc.SOAPFaultException, rather than a service-specific exception, when implementing a WebLogic Web Service.

XML Known Issues

Change Request Number

Description

CR102744

Although the class weblogic.xml.xpath.XMLNodeXPath is not public, its Javadoc was mistakenly exposed. BEA does not support the use of this internal class.

 

Skip navigation bar  Back to Top Previous Next