This document contains the release notes for IBM WebSphere Application
Server, Version 5.0.1. For the very latest updates, view the English online
version of the Release Notes located at http://www.ibm.com/software/webservers/appserv/library/
.
Installation and uninstallation
Cannot install the pluggable client from the IBM
WebSphere Application Server client CD on a Windows system
You cannot install the pluggable client from the IBM WebSphere
Application Server client CD on a Windows system.
To
work around this problem, do one of the following:
- Copy the IBM WebSphere Application Server client CD for Windows installation
image to your local temp directory and install
it from there.
- Run the installer manually from a command line with the following options:
<cd drive>\nt\install.exe -P OrbPropertiesFileCopy.installLocation="<SUNJrePath>/lib"
Where <cd
drive> is the CD drive and <SUNJrePath> is the location
of the SUN Operating Environment JRE file installed on your machine. For example:
F:\nt\install.exe -P OrbPropertiesFileCopy.installLocation="C:/JavaSoft/JRE/1.3.1_04/lib"
VersionInfo command
parameters are incorrectly described
The description of the versionInfo command
is incorrect in the readme_updateinstaller files included with the
Fix Pack 1 files.
Parameters do not include display.
Use
these parameters to create a version report:
- -format
- TEXT | HTML
Selects the format of the report. The default
is TEXT.
- -file
- <fileName>
Specifies the output file name. The
report goes to standard output (stdout) by default.
- -long
- Produces the long version of the report.
- -components
- Adds a list of installed components to the report.
- -componentDetail
- Adds details about installed components to the
report.
- -efixes
- Adds a list of applied fixes to the report.
- -efixDetail
- Adds details about applied fixes to the report.
- -ptfs
- Adds a list of applied fix packs to the report.
- -ptfDetail
- Adds details about applied fix packs to the report.
If you enter an incorrect parameter, the versionInfo command
displays an error message that describes the parameter as invalid.
Use
the -? or -help parameter to display correct command syntax.
Save the jdbc-resource-provider-templates.xml file
before you install the WebSphere Application Server, Version 5 Fix Pack 1
In
WebSphere Application Server, Version 5.0.1, the jdbc-resource-provider-templates.xml file
is modified to include a new DB2 Universal JDBC provider and other JDBC provider
name changes. If you have modified this configuration file in WebSphere Application
Server, Version 5, save the jdbc-resource-provider-templates.xml file
before installing the WebSphere Application Server, Version 5.0.1. Otherwise,
the new XML file overwrites the jdbc-resource-provider-templates.xml file
and your configuration data is lost.
Extra information
displays in the Network Deployment response file
Extra information
displays in the IBM WebSphere Application Server Network Deployment response
file. Specifically, the following line appears and always must be commented:
#-W coexistencePanelBean.useIhs="true"
Port number used by the wsadmin tool can change
during coexistence scenario
During a coexistence scenario, you
can change the port number used by the wsadmin tool. If you do change the
port number, you must invoke the wsadmin tool using the new port using one
of the following commands:
wsadmin.bat/sh -conntype RMI -port <port_number>
wsadmin.bat/sh -conntype SOAP -port <port_number>
Alternatively,
you can update the wsadmin.properties file located in the <WAS_INSTALL_ROOT>/properties directory
to contain the new port values. If you do this, you do not need to specify
the port parameter when using the wsadmin tool.
Installation requirements
for the embedded messaging component
If you want to install the embedded messaging component on IBM
WebSphere Application Server and IBM WebSphere Application Server Network
Deployment on the same machine for the Windows NT and Windows 2000 platforms,
you must install IBM WebSphere Application Server first and then install IBM
WebSphere Application Server Network Deployment. Otherwise, embedded messaging
installation fails.
Coexistence problem between embedded
messaging, IBM WebSphere Studio Application Developer Integration Edition, and IBM WebSphere Application Server
The
IBM WebSphere Studio Application Developer Integration Edition and IBM WebSphere
Application Server both include an option to install embedded messaging. The
embedded messaging option in these two products is incompatible.
To
avoid this problem, do not install embedded messaging for both products on
the same machine.
When installing IBM
WebSphere Application Server Enterprise, Version 5 on top of the Deployment
Manager, the Javadoc is missing
When you install the IBM WebSphere
Application Server Enterprise, Version 5 on top of the Deployment Manager
from IBM WebSphere Application Server Network Deployment, Version 5 , the
Javadoc for the IBM WebSphere Application Server Enterprise, Version 5 is
missing.
To work around this problem, see the IBM WebSphere Application
Server Support Web site at the following address: http://www-3.ibm.com/software/webservers/appserv/support/index.html
System calls are not restarted correctly within
the Java virtual machine
A
problem exists in the Sun Java Development Kit on the Solaris Operating Environment,
which is fixed in 1.3.1_07 (FCS Jan 2003), when the socket reads and writes
randomly thrown InterruptedIOException or java.net.SocketException exceptions.
The root problem is that system calls are not restarting as they should within
the Java virtual machine.
See bugs numbers 4425033 and 4178050 at http://developer.java.sun.com/developer/bugParade
for
further details. If previous exceptions are encountered and are having a significant
impact on your operation before availability of 1.3.1_07, contact IBM for
support.
Preparing to uninstall WebSphere Application Server,
Version 5 Fix Pack 1 (with WebSphere Embedded Messaging) on Solaris
Before uninstalling WebSphere Application
Server, Version Fix Pack 1 on Solaris, complete the following steps:
- If you have a WebSphere Application Server, or WebSphere Application Server
Network Deployment installation that includes the WebSphere Embedded Messaging
Client without the Server component, complete the following steps, before
uninstaling the Version 5 Fix Pack 1 on Solaris.
Note: If you installed
WebSphere Application Server with both the WebSphere Embedded Messaging Client
and Server components, you can ignore this item because it does not apply
to your installation.
The required procedure varies slightly depending
on how you chose to install the embedded WebSphere Embedded Messaging components:
- WebSphere Application Server and Embedded Messaging Client (only) installed
- Locate and delete the mqVer.properties file in the /opt/WebSphere/DeploymentManager/properties/version/backup/external.mq/ptfs/was50_nd_fp1_solaris/components/external.mq directory.
- Manually uninstall the WebSphere embedded messaging service -/usr/sbin/pkgrm
mqm-upd03.
- Continue with uninstall of WebSphere Application Server, Version 5 Fix
Pack 1.
- WebSphere Application Server Network Deployment and Embedded Messaging
Client (only) installed
- Locate and delete the mqVer.properties file in the /opt/WebSphere/AppServer/properties/version/backup/external.mq/ptfs/was50_fp1_solaris/components/external.mq directory.
- Manually uninstall the WebSphere embedded messaging service - /usr/sbin/pkgrm
mqm-upd03.
- Continue with uninstall of Version 5 Fix Pack 1.
If you do not complete this procedure, when you uninstall the Version
5 Fix Pack 1 on Solaris, the following error messages are added to the fix
pack uninstallation log. To recover from these errors, complete the procedure,
then repeat the fix pack uninstallation.
PTF Component Activity:
=========================================================
Component Name : external.mq
Action : uninstall
Time Stamp (Start) : 2003-04-09T10:41:12-05:00
=========================================================
Log File Name : /opt/WebSphere/DeploymentManager/properties/version/log/20030
409_154112_was50_nd_fp1_solaris_external.mq_uninstall.log
Backup File Name : /opt/WebSphere/DeploymentManager/properties/version/backup/20
030409_154112_was50_nd_fp1_solaris_external.mq_undo.jar
=========================================================
Saving History ...
Saving History ... Done
Running component update ( external.mq ).
Processing MQ CSD Uninstall Script
Processing MQ CSD version information.
Performing CSD uninstall.
Failed to perform uninstall.
Exception: WUPD0246E: Fix pack update failure: An exception occurred while
preprocessing the content of fix pack was50_nd_fp1_solaris, component external.mq
Saving History ...
Saving History ... Done
Results:
=========================================================
Time Stamp (End) : 2003-04-09T10:41:14-05:00
PTF Component Result : failed
PTF Component Result Message:
=========================================================
WUPD0246E: Fix pack update failure: An exception occurred while preprocessing the
content of fix pack was50_nd_fp1_solaris, component external.mq
=========================================================
PTF Component Installation ... Done
Exception: WUPD0243E: Fix pack uninstall failure: The update for component {1} for
fix pack external.mq could not be uninstalled.
IBM Toolbox for Java JDBC driver
WebSphere Application
Server supports the IBM Toolbox for Java JDBC driver. The IBM Toolbox for
Java JDBC driver is included with the IBM Toolbox for Java product.
IBM
Toolbox for Java is a library of Java classes that are optimized for accessing
iSeries and AS/400 data and resources. You can use the IBM Toolbox for Java
JDBC driver to access local or remote DB2 UDB for iSeries 400 databases from
server-side and client Java applications that run on any platform that supports
Java.
IBM Toolbox for Java is available in these versions:
- IBM Toolbox for Java licensed program
The licensed program is available
with every OS/400 release, Version 4 Release 2 (V4R2) or later. You can install
the licensed program on your iSeries 400 system, and then either copy the
IBM Toolbox for Java JAR file (jt400.jar) to your system or update
your system classpath to locate the server installation. Product documentation
for IBM Toolbox for Java is available from the iSeries Information Center:
http://publib.boulder.ibm.com/pubs/html/as400/infocenter.html (The documentation is located under Programming > Java > IBM Toolbox
for Java.)
- JTOpen
JTOpen is the open source version of IBM Toolbox for Java, and
is more frequently updated than the licensed program version. You can download
JTOpen from http://www-1.ibm.com/servers/eserver/iseries/toolbox/downloads.htm. Also available for download is the JTOpen Programming Guide. The guide
includes instructions for installing JTOpen, as well as information about
the JDBC driver.
The JDBC driver for both versions supports JDBC 2.0. For more information
about IBM Toolbox for Java and JTOpen, see the product Web site at http://www-1.ibm.com/servers/eserver/iseries/toolbox/index.html.
For customers using WebSphere Application Server, Version 5 on platforms
other than iSeries, the JTOpen version of the Toolbox JDBC driver is the one
to download and use.
To configure or use the jt400.jar file,
do the following:
- Download from the jt400.jar file from the JTOpen link. Place
it in a directory on your work station such as C:\JDBC_Drivers\Toolbox.
- In the WebSphere Application Server, Version 5 administrative console,
use the following links to set the WebSphere Application Server managed variable OS400_TOOLBOX_JDBC_DRIVER_PATH at
the Node level: Environment > Managed WebSphere Variables.
- Double-click OS400_TOOLBOX_JDBC_DRIVER_PATH.
- Set the value to the full directory path to the jt400.jar file
downloaded in step one. Do not include the jt400.jar file in this
value. For example, OS400_TOOLBOX_JDBC_DRIVER_PATH == "C:\JDBC_Drivers\Toolbox"
This path is referenced in the jdbc-resource-provider-templates.xml file
used to configure JDBC resource providers.
Complete the following
steps to choose a Toolbox driver from the list of possible resource providers:
- Click Resources > JDBC Providers.
- Select New.
- From the JDBC Provider list, click UDB DB2 for iSeries (Toolbox) or UDB
DB2 for iSeries (Toolbox XA).
When the provider is created, the Classpath field looks as follows: Classpath
= ${OS400_TOOLBOX_JDBC_DRIVER_PATH}/jt400.jar.
Installing IBM HTTP Server updates
with WebSphere Application Server Network Deployment Fix Pack 1
Fix
Pack 1 for WebSphere Application Server Network Deployment contains updates
for IBM HTTP Server, which allows you to update a copy of IBM HTTP Server
that is installed on the same machine as the deployment manager. If the same
machine has the deployment manager and the base WebSphere Application Server
with as IBM HTTP Server as a feature, you only need to update IBM HTTP Server
with the fix pack for base or the fix pack for Network Deployment but not
both. To see if IBM HTTP Server has been updated, check the version.signature file
in the IBM HTTP Server installation root directory. If the file says "IBM
HTTP Server 1.3.26.1", then Fix Pack 1 is applied to IBM HTTP Server.
Uninstall fixes for IBM HTTP Server and MQ Series
(Embedded Messaging) before installing fix pack updates to these features
If
you have installed fixes for IBM HTTP Server or MQ Series Embedded Messaging,
you need to uninstall these fixes prior to updating these features using the
Update Installer. If you do not uninstall these fixes, the updates to IBM
HTTP Server or Embedded Messaging made during the fix pack installation might
fail or the installation might be faulty. You can choose to skip the updates
IBM HTTP Server or Embedded Messaging if they are not required during fix
pack installation. See the Update Installer documentation for more information.
Port updates for co-existence require a WebSphere
Application Server installation
Port updates for co-existence require
the installation of WebSphere Application Server. This requirement affects
port updates for IBM HTTP Server co-existence. Port updates do not occur if
only the IBM HTTP Server is installed. In this case, manually update the httpd.conf files.
The administrative console, or the adminconsole.ear file,
does not function properly when com.ibm.websphere.sendredirect.compatibility is
set to true
If you need to create a Java virtual machine
(JVM) com.ibm.websphere.sendredirect.compatibility custom property
and set it to true, create an additional server and install the application
on the new server. The administrative application does not function properly
when the com.ibm.websphere.sendredirect.compatibility property is set to true.
The administrative application fails because the login URL is modified
by adding a second /admin.
To work around this problem, delete
the second /admin from the URL in the browser.
Using the correct working directory
when you run installation or uninstallation
When you run installation
or uninstallation, the current working directory must be the directory where
the installer or uninstaller binary is located. This placement is important
because the resolution of the class files location is done in the current
working directory.
For example:
cd INSTALL_ROOT/_uninst
./install
or:
cd <CD mount point>
./install
Failure to use the correct working directory can cause
the ISMP errors that abort an installation or uninstallation.
You must restart the server after a configuration
change
If you make any changes to the configuration, restart the
server as noted in the messages section of the administrative console.
A core dump might occur when running WebSphere
Application Server with DB2 V7.2 FP8 client in AIX 5.2
A core dump might occur when you are running WebSphere Application
Server with DB2 V7.2 FP8 client in AIX 5.2 having all of the following configurations:
- You installed WebSphere Application Server in 64-bit AIX 5.2.
- You installed 64-bit DB2 V7.2 FP8.
- You created a 32-bit client instance to connect to the DB2 server.
- You installed applications which use DB2 data source in WebSphere Application
Server.
A java core dumped error displays when you start the WebSphere
Application Server. This is due to a DB2 problem. A defect has been opened
to DB2.
To solve this problem, rename the libdb2lai.a file
on your DB2 client machine as follows:
- Change to the /usr/lpp/db2_07_01/lib directory.
- Rename the file libdb2lai.a to libdb2lai.a.orig.
Migration
Changes are made to the Java Database Connectivity
providers in WebSphere Application Server Fix Pack 1
The following
changes are made to the Java Database Connectivity (JDBC) providers in WebSphere
Application Server Fix Pack 1:
- Fix Pack 1 provides a new JDBC provider for DB2. Use this JDBC provider
to test the new DB2 Universal JDBC type 4 driver. This provider only supports
5.0 datas ource in WebSphere Application Server, Version 5.0.1. To distinguish
between this new jdbc provider and the existing DB2 type 2 jdbc provider,
Fix Pack 1 changes the name of the provider in thejdbc-resource-provider-templates.xml file
to match the provider with the name.
- Fix Pack 1 merges both Oracle thin and oci drivers into one single Oracle
provider because there are no property differences between these two drivers.
Fix Pack 1 also marks the old jdbc providers as deprecated.
- The old Cloudscape and Sybase JDBC providers have version numbers in the
names. To support new versions under the same JDBC driver provider in the
future, Fix Pack 1 removes the version numbers. Fix Pack 1 also adds two Cloudscape
and Sybase JDBC providers. Fix Pack 1 marks the old Cloudscape and Sybase
JDBC providers as deprecated.
The removeNode
command removes clusters from configuration
When migrating from
a previous IBM WebSphere Application Server version to IBM WebSphere Application
Server Network Deployment, you must first migrate each of the nodes and then
use the addNode command to federate the WebSphere base product nodes
into the Network Deployment environment. Later, the removeNode command
can be used to remove the WebSphere base node from the Network Deployment
environment.
But, the removeNode command removes the configuration
entries from the clusters.xml file for the node. When you attempt
to install an application on the cluster no members are found because the removeNode command
deleted the entries.
To work around this problem, if you execute the removeNode command
on a node that contains servers that are part of a cluster, you need to explicitly
create new cluster member servers on that node if you want to add the node
to the cell again.
Note:There are no functional changes, Fix Pack
1 only changes the names for usability reason.
The changed providers are:
| Data source |
JDBC providers in Version 5 |
JDBC providers in Version 5.0.1 |
| DB2 |
DB2 JDBC provider (deprecated) |
DB2 legacy CLI-based type 2 JDBC driver |
| |
DB2 JDBC provider (XA )(deprecated) |
DB2 legacy CLI-based type 2 JDBC driver (XA) |
| |
|
DB2 Universal JDBC driver provider (type 4 only) |
| Oracle |
Oracle JDBC thin driver (deprecated) |
Oracle JDBC driver |
| |
Oracle JDBC thin driver (XA) (deprecated) |
Oracle JDBC driver (XA) |
| |
Oracle JDBC oci8 driver (deprecated) |
Oracle JDBC driver |
| |
Oracle JDBC oci8 driver (XA) (deprecated) |
Oracle JDBC driver (XA) |
| Cloudscape |
Cloudscape JDBC provider 5.0 (deprecated) |
Cloudscape JDBC provider |
| |
Cloudscape JDBC provider 5.0 (XA) (deprecated) |
Cloudscape JDBC provider (XA) |
| Sybase |
Sybase 12.0 JDBC driver (deprecated) |
Sybase JDBC driver |
| |
Sybase 12.0 JDBC driver (XA) (deprecated) |
Sybase JDBC driver (XA) |
For a complete description of supported JDBC driver providers, required
JDBC driver files and data source properties, see the Vendor specifications
file (rdat_minreq) in the InfoCenter.
What do these changes mean to you?
If you use the Java command language Jacl to create a JDBC provider and
data source and your script references those names (for example, you use the
createUsingTemplate API to create a jdbc provider), you might have to change
your Jacl to use the new name. Since the old names are deprecated, we recommend
you change the names as soon as possible.
XMLConfig utility and migration integration
Migration
uses the XMLConfig utility to save the WebSphere Application Server Advanced
Edition, Version 4.0 configuration as part of migration. There are some fixes
available to the IBM WebSphere Application Server Advanced Edition, Version
4.0 XMLConfig utility that you can apply if problems are occurring during
the call to this utility during migration including:
PQ52555 - XMLConfig does not export clone property configuration
PQ55064 - XMLConfig does not export enterprise bean to DataSource level mappings
PQ58038 - Performing an XMLConfig export produces an extra CRLF
PQ62103 - XMLConfig full export fails with NullPointerException in a Multi-Node Environment
PQ62471 - Security AdminRoles are not getting exported during XML export
PQ63815 - "=" is not a valid character for the value string in XMLConfig
The WASPostUpgrade
migration tool fails
You might receive an error in either of the
following situations:
- The IBM WebSphere Application Server migration tools are run manually
in an IBM WebSphere Application Server Network Deployment environment in which
there are Network Deployment nodes. These migration tools are provided to
move configurations from previous versions of the IBM WebSphere Application
Server.
- The IBM WebSphere Application Server Enterprise is installed in an IBM
WebSphere Application Server Network Deployment environment in which there
are Network Deployment nodes.
The errors, which display either on the command-line or in the
Migration panel of the installation program, will be similar to the following:
MIGR0122E: Unable to read configuration file /opt/WebSphere/DeploymentManager/config/cells/
ws-sunfish2Network/nodes/
at com.ibm.websphere.migration.postupgrade.TransformBaseConfiguration.resolveDirectory
(TransformBaseConfiguration.java:690)
at com.ibm.websphere.migration.postupgrade.TransformBaseConfiguration.initializeDataModel
(TransformBaseConfiguration.java:587)
at com.ibm.websphere.migration.postupgrade.TransformBaseConfiguration.doIt
(TransformBaseConfiguration.java:1145)
at com.ibm.websphere.migration.postupgrade.Restore.doIt(Restore.java:153)
at com.ibm.websphere.migration.postupgrade.Restore.<init>(Restore.java:93)
at com.ibm.websphere.migration.postupgrade.WASPostUpgrade.restore(WASPostUpgrade.java:196)
at com.ibm.websphere.migration.postupgrade.WASPostUpgrade.main(WASPostUpgrade.java:665)
at java.lang.reflect.Method.invoke(Native Method)
at com.ibm.ws.bootstrap.WSLauncher.main(WSLauncher.java:94)
com.ibm.websphere.migration.exceptions.WASUpgradeInvalidConfigurationException:
MIGR0122E: Unable to read configuration file /opt/WebSphere/DeploymentManager/config/cells/
ws-sunfish2Network/nodes/.
To work around this problem, invoke the WASPostUpgrade migration
tool after installation is complete. You must specify the -nodeName parameter
with the deployment manager node when you use the WASPostUpgrade command.
For more information on the WASPostUpgrade command, see the article
entitled, "WASPostUpgrade command " in the IBM WebSphere Application Server,
Version 5 documentation.
The maxSendSize and maxReceiveSize variable
values do not migrate from IBM WebSphere Application Server, Version 4 to
Version 5
After migrating to IBM WebSphere Application Server, Version
5, the maxSendSize and maxReceiveSize variable values, which are included
in the WorkArea Service, are set to the Version 5 default values. These variables
are not correctly mapped from Version 4 to Version 5.
To work around
this problem, use the Customer Services section of the administrative
console to update these variables with their Version 4 values.
NullPointerExeption
thrown when you click the Test Connection for a 4.0 data source that is created
from the WASPostUpgrade path
When you click Test Connection for
a 4.0 data source that is created from the WASPostUpgrade path, the
following exception displays:
DSRA8040I: Failed to connect to the Data Source. Encountered: java.lang.NullPointerException
This
exception is caused by a 4.0 data source that is created through the WASPostUpgrade path
does not have any custom properties.
To avoid receiving the exception,
create a valid DB2 custom property on the data source. For example, create
a description in the Custom Property panel.
System administration
System performance degrades when you configure
the nodeagent failover with a certain script
If you configure the nodeagent failover
with the script created by {WAS_HOME}/bin/startServer.sh nodeagent
-script startnodeagent.sh -background script, the system performance
degrades. A message similar to the following displays:
Feb 28 14:48:37 wastst12 init: Id "was" respawning too fast: disabled for 5 minutes
Configure the nodeagent failover with the script created by {WAS_HOME}/bin/startServer.sh
nodeagent -script startnodeagent.sh to work around the problem.
Cannot use a server
name that is a member of one cluster in another cluster
You cannot
use a server name assigned to a cluster member elsewhere in the cell. For
example, Cluster1 can have multiple members named server1, but no server in
the cell outside the cluster can have the name server1. Otherwise, lookups
which resolve to an object on a cluster, or a single server involved in the
name conflict can fail. To work around the problem, assign unique names to
cluster members.
Plugin-cfg.xml
files on federated nodes can be overwritten
Because the
plugin-cfg.xml file
on a Network Deployment machine is stored in the configuration repository,
the
plugin-cfg.xml file is overwritten on all federated nodes in
your network whenever the Network Deployment machine updates the configuration
repository on these federated nodes. If you customize the
plugin-cfg.xml files
on your federated nodes, change the name of
plugin_cfg_custom.xml and
httpd.conf files
in the Web server to point to the new file name. In a Windows environment,
you must edit the Windows NT registry to update the plugin-cfg key to point
to the new
plugin-cfg.xml file for Domino and Internet Information
Services (IIS) Web servers.
Issuing the addNode command
can replace the plugin-cfg.xml file
After
federating the Node, the addNode command backs up the plugin-cfg.xml file
from the <install_root>/config/cells directory to
the config/backup/base/cells directory. The addNode command
in Fix Pack 1 regenerates a new plugin-cfg.xml file at the Deployment
Manger and the nodeSync operation copies the files to the node level. You
might experience a Web server failure due to the wrong directory name in the
new plugin-cfg.xml file.
Wrong directory
name in the Network Deployment plugin-cfg.xml file
When
you regenerate the plugin-cfg.xml file through the network deployment
manager, the resulting file contains path information specific to the Network
Deployment installation. This file is replicated to the node level. Or, you
can copy this file to a remote Web server. If the directory names do not
match the product installation directory, the Web server cannot start. Edit
the plugin-cfg.xml file and replace all occurrences of network_deployment_install_root with
websphere_application_server_install_root.
For example, if
you installed IBM WebSphere Application Server into the directory C:\WebSphere\DeploymentManager,
replace all occurrences of C:\WebSphere\DeploymentManager with C:\WebSphere\AppServer.
Use any text editor to change and edit any file locations.
For exmaple,
the plugin-cfg.xml file has a line:
<Log LogLevel="Error" Name="/opt/WebSphere/DeploymentManager/logs/http_plugin.log"/>
which needs to change to
<Log LogLevel="Error" Name="/opt/WebSphere/AppServer/logs/http_plugin.log"/>
WebSphere
Application Server Version 5 Fix Pack 1 has a new feature to retain the directory
names you have changed in the plugin-cfg.xml file. If you want to
regenerate the new plugin-cfg.xml file with the default values, delete
the current plugin-cfg.xml file before generating the new file.
The -stopservers option
on the syncNode utility does not work correctly
The -stopservers option
on the syncNode utility does not work correctly if the node agent server
is running. The syncNode tool is not needed if the node agent is running.
The node agent
autoRestart feature depends on functional multicast protocol
The
node agent monitors running application servers and other managed processes,
and automatically restarts the process if it crashes or becomes non-responsive.
This feature depends on the multicast IP protocol functioning between the
node agent and the managed processes. If there is a problem with multicast
protocol functioning on the system, or if it is desired to disable multicast
support in the subnet where the system is running, the node agent autoRestart
feature does not work correctly.
Servers can restart by mistake if multicast
protocol is not working. In this situation, it is recommended to disable node
agent autoRestart for the affected servers, and rely on the native operating
system monitoring feature instead. The native operating system monitoring
involves configuring the server as a Windows service on Windows operating
systems, or creating an inittab entry for the rc.was script
on UNIX operating systems. Information on setting up native operating system
monitoring is in the documentation.
Unsynchronized
system clocks cause failure
The security system requires system
clocks across a secure cell to be synchronized. When running in a secure Network
Deployment environment and the clocks of the systems in the cell are not synchronized,
administrative operations fail because of authorization failures due to expired
LTPA tokens.
See the article entitled, "Security considerations specific
to a multi-node or process Network Deployment environment" for more information.
Empty NetworkDeploymentCell
directory is left in the configuration directory
When you install
IBM WebSphere Application Server, and you run addNode then removeNode, sometimes
an empty NetworkDeploymentCell directory is left in the configuration directory
making wsadmin think there are two cells configured.
The WebSphere product
cannot delete the directory because the directory is locked by another application.
The most common reason for this situation is that another application uses
that directory or its subdirectory as the current working directory, or another
application opens a file under that directory or subdirectory. This problem
is most commonly seen on Windows platforms, but it can happen on other platforms.
To
work around this problem, manually delete the empty directory structure, then
restart the servers that use this configuration directory.
Accessing First Steps items on Linux 390 systems
When you select the following features from
the First Steps GUI, the Web browser window does not open:
- WebSphere InfoCenter
- Register the Product
- Samples Gallery
- Administrative Console
To access these features:
- Open a Web browser window on another machine.
- Type the browser URL that displays in the First Steps status window into
the address field of the Web browser.
- Click Enter to open the page in the Web browser.
The Domino Server plug-in fails to configure on
Solaris Operating Environment
During installation, a dsapi_stderr.txt file
is created in the logs directory and you can get
the following error messages:
lotus.notes.NotesException: Could not load dll for system name SUNOS
at lotus.notes.NotesThread.load(NotesThread.java:210)
at lotus.notes.NotesThread.<clinit>(NotesThread.java:24)
java.lang.UnsatisfiedLinkError: NnotesInitThread
at lotus.notes.NotesThread.NnotesInitThread(Native Method)
at lotus.notes.NotesThread.initThread(NotesThread.java:99)
at lotus.notes.NotesThread.run(NotesThread.java:133)You
can configure the IBM WebSphere Application Server or Domino Server plug-in
manually using the Domino Server Web administration tool. The workarounds
are as follows:
- Start the Domino Server.
- Enter the URL for the Domino Server Web Administration site using a browser.
For example, http://<hostname>/names.nsf.
Enter the administrator user name and password.
- Double-click Server-Servers.
- Double-click WebServer to configure.
- Double-click Edit Server.
- Double-click Internet Protocol.
- Add the IBM WebSphere Application Server DSAPI plug-in to the DSAPI field.
For example, /opt/WebSphere/AppServer/bin/libdomino5_http.so
Note: If
there are already DSAPI filter files specified, use a space to delimit the
WebSphere plug-in file.
- Double-click Save and Close.
- Restart the Domino Server.
Cluster state
not always updated when using Windows Services
When starting and stopping application servers that are part of
a cluster using the Windows Services facility, the cluster state does not
always get updated correctly. For example, when a cluster is running and you
stop a cluster member through the Services GUI, the cluster state remains
as "Started" even though the server is no longer running.
Enabling authentication in
the file transfer service
In WebSphere Application Server Network
Deployment, Version 5.0.1, the file transfer service is enhanced to provide
role-based authentication. Two versions of the file transfer Web application
are provided with this fix pack. By default, the version that does not authenticate
its caller is installed. This default allows for compatibility between the
WebSphere Application Server Network Deployment, Version 5 and Version 5.0.1.
After
all the nodes in your cell are upgraded to WebSphere Application Server Network
Deployment, Version 5.0.1, you can activate the authentication in the file
transfer by redeploying the file transfer application at the deployment manager.
The compatible version is shipped in the file ${INSTALL_ROOT}/installableApps/filetransfer.ear file.
The secured version is provided in the ${INSTALL_ROOT}/installableApps/filetransferSecured.ear file.
A wsadmin Jacl
script is provided to help you redeploy the file transfer. The script is redeployFileTransfer.jacl and
can be found in the ${INSTALL_ROOT}/bin directory. After the deployment
manager and all the nodes have been upgraded to WebSphere Application Server
Network Deployment, Version 5.0.1, you can deploy the secured file transfer
service by running the script. The syntax for running the script is:
wsadmin -profile redeployFileTransfer.jacl -c "fileTransferAuthenticationOn ear_file_path"
For example, when executing the script from the c:/WebSphere/DeploymentManager/bin directory,
the syntax is similar to the following:
wsadmin -profile redeployFileTransfer.jacl -c "fileTransferAuthenticationOn c:/WebSphere/DeploymentManager/installableApps"
If you want to go back to run the file transfer service without authentication,
you can run the script as follows:
wsadmin -profile redeployFileTransfer.jacl -c "fileTransferAuthenticationOff ear_file_path"
Administrative console
All platforms
Administrative Console - Netscape browser
All
platforms
Application servers must
have different disk offload locations
When you have two or more
application servers with servlet caching enabled and the application servers
specify the same disk offload location for their caches through the dynamic
cache service, the following exceptions might occur:
java.lang.NullPointerException
at com.ibm.ws.cache.CacheOnDisk.readTemplate(CacheOnDisk.java:686)
at com.ibm.ws.cache.Cache.internalInvalidateByTemplate(Cache.java:828)or:
java.lang.NullPointerException
at com.ibm.ws.cache.CacheOnDisk.readCacheEntry(CacheOnDisk.java:600)
at com.ibm.ws.cache.Cache.getCacheEntry(Cache.java:341)If
one server is run as root and the other servers are run as nonroot, this problem
could occur. For example, if server1 runs as root and server2 runs
as wasuser or wasgroup, the cache files in the disk offload
location might be created with root permissions. This situation causes the
applications running on the nonroot servers to crash when they try to read
or write to the cache.
The disk offload location must be unique for
servers defined on the same node. If you have multiple servers defined on
the same node, make sure the disk offload location is different for each server
as defined on the Dynamic Cache Service panel, Offload location
field.
An Error 500 message
displays when you click Name Space Bindings
On UNIX platforms, when you click Name Space Bindings,
you receive an Error 500 message on the browser. When you click Show Details,
a NullPointerException displays. This only occurs with the administrative
ID that was configured as the server ID.
To solve the problem, add a
different user to System Management > Console Users and login with
that user to add additional Name Space Bindings.
Adding a secured
remote node through the administrative console
Adding a secured
remote node through the administrative console is not supported. You can either
disable security on the remote node before performing the operation or perform
the operation from the command line using the addNode script.
Administrative console does not support browser
Back and Forward mechanisms
The administrative console does not
support using the browser Back and Forward mechanisms. Use of
these mechanisms can cause intermittent problems. Use Back or Cancel on
the administrative console panels instead.
Enabling the cookies in the browser
You
must enable the cookies in the browser to have the administrative console
work correctly.
Do not change the name of the JDBC provider after
it is created
When using the administrative console, do not change
the name of the JDBC provider if you create it by selecting an existing JDBC
provider from the menu. If you use a custom JDBC provider, for which you supply
the provider information, this restriction does not apply.
Error occurs when
accessing the Cluster topology panel
If your first action after
logging into the console is to click Servers > Cluster Topology and
then click on a link to a cluster, the message ServletException in:/secure/layouts/detailTitleLayout.jsp]
null displays at the top of the panel instead of the bread crumb links
and the panel description.
To work around this problem, access any other
panel in the administrative console first and then access the Cluster topology
panel.
Resolve the administrative console port conflict
If
the administrative console port is in conflict with an application already
running on the machine, you can change this port in the following files:
- <install_root>/config/cells/<cellname>/nodes/<nodename>/servers/<servername>/server.xml
- <install_root>/config/cells/<cellname>/virtualhosts.xml
Change all occurrences of port 9090 (or whatever port was selected during
installation time in an IBM WebSphere Application Server coexistence scenario)
to the desired port for the administrative console.
Alternatively, shut
down the other application using this port before starting the IBM WebSphere
Application Server.
The addNode command
does not remove the administrative console transport port
If you
configured the IBM WebSphere Application Server administrative console transport
port to a setting other than the 9090 default, federating the node using the addNode command
does not remove this nondefault transport from the server1 configuration.
If you use the same transport number for the IBM WebSphere Application Server
Network Deployment administrative console, you cannot start the application
server because of a port conflict. You can remove this unused transport port
using the IBM WebSphere Application Server Network Deployment administrative
console before starting server1.
The test connection reports
errors if an environment variable in the definition of a data source is used
If
you use an environment variable in the definition of a data source, for example, classpath=${DB2_JDBC_DRIVER_PATH}/db2java.zip,
the test connection reports an error message similar to the following:
DSRA8040I: Failed to connect to the DataSource.
Encountered : java.lang.ClassNotFoundException: DSRA8000E: No jar or zip
files found in /db2java.zip
To test a connection, define the WebSphere environment variable DB2_JDBC_DRIVER_PATH at
the manager level regardless of the level at which the data source is defined.
The normal scoping rules for the variables are not in effect for the test
connection.
Resizing the Netscape browser results in an error
If
you resize your Netscape browser, you could get a "Data Missing" error.
The
error message disappears in 60 seconds.
Enabling Netscape Version 4.7 to display
double-byte character set correctly
When using Netscape on AIX platforms without the translated package,
the English version of Netscape is available for all locale environments as
the default package. However, the English version of Netscape does not display
double-byte character set (DBCS) characters on the browser radio buttons and
title bars because the fonts are mismatched.
To work around this problem,
you can install the message resource to make the translated version of Netscape
available on a DBCS environment. Use the translated version of Netscape to
display the corrupted DBCS correctly. Change the locale from English to the
expected DBCS locale before starting Netscape. For example, issue the following
commands to display the Japanese contents on Ja_JP (AIX Shift JIS locale):
$ export LANG=Ja_JP
$ netscape&
How to check the installed package
For
the AIX platform:
- Execute the lslpp -aL | grep Netscape.msg command.
- Check the language and locale information using Netscape.msg package.
For example, the result of the command executed in step 1 is as follows for
the system having Japanese-translated Netscape:
# lslpp -aL | grep Netscape.msg
Netscape.msg.Ja_JP.communicator.rte
Netscape.msg.ja_JP.communicator.rte
For the Red Hat Linux Version 7.2 platform:
Execute the rpm-ql
netscape-common command to get the list of installed files.
In case
of Netscape Version 4.78 on Red Hat Linux Version 7.2, the translated packages
for seven languages including four DBCS languages (Japanese, Korean, Simplified
Chinese, and Traditional Chinese) are included in a common package.
How
to install the package
For the AIX platform:
Install the Netscape.msg.%L.communicator.rate package where %L is
the locale information for the System Management Interface Tool (SMIT) or
SMITTY to use. This message resource is provided by the Bonus Pack CD or the
Expansion Pack CD for AIX platform, which provides the Netscape common package.
Perform
the following steps to install the package:
- Set the media which has the Netscape installation image. For AIX 5L platform,
set the Expansion Pack media.
- Go to the directory which has the Netscape installation package. For AIX
5L platform Expansion Pack, go to the installp/ppc directory.
- Start SMIT or SMITTY.
- Click System Environments > Manage Language Environment > Change/Show
Applications for a Language > Add Language for Application Already
Installed.
- In the Add Language for Application Already Installed field select
a language (locale) from the drop-down menu list for the Language translation
to install field.
- Click OK.
- In the Add Language for Application Already Installed field enter
the directory location. Enter "." for the current directory or /cdrom/installp/ppc for
the full path as input directory for the software. Do not select cdrom as
the input device because the installation package does not exist on the root
directory of the CD-ROM.
- Click OK.
- Click List to check if the configuration is correct. The configuration
is correct if Netscape is on the list of installed application. Execute the
installation now.
- Click OK.
- Exit SMIT or SMITTY.
For Red Hat Linux Version 7.2 platform:
The default Netscape
installation can install each translated resource as part of the common package.
There is no way to install the translated package individually. However, TrueType
fonts (TTF packages) require installation for each language to correctly display
the translated resources on Netscape. Use the following to check if the required
font files are correctly installed:
$ rpm -qa | grep ttf
resources.0-4
ttfm-0.9.1-8
ttfonts-ja-1.0-6
ttfonts-ko-1.0-10
ttfonts-zh_CN-2.11-5
ttfonts-zh_TW-2.11-5
Resizing the Netscape browser causes an error
404 message
When connecting to the IBM WebSphere Application
Server administrative console from a Netscape browser, resizing the browser
can cause an error 404 message to occur. This
situation occurs because the browser reloads the frame when resizing the window.
To
avoid getting the error message, you can refrain from resizing the Netscape
browser window, or you can connect to the IBM WebSphere Application Server
administrative console using an Internet Explorer browser.
Using Netscape Version 4.79 on a Solaris Operating
Environment causes problems
Using
Netscape Version 4.79 on a Solaris Operating Environment to access the administrative
console causes problems with some key text translations with the zh_TW.EUC locale.
This situation is not a problem when you use Netscape Version 4.7. The officially
supported version of Netscape on a Solaris Operating Environment is Version
4.79, but in this case the workaround is to use Netscape Version 4.7.
Limitations occur when using Netscape with Solaris
Operating Environment
If
you click Troubleshooting > Logs and Trace > <server> > Diagnostic
Trace > Modify, the window that pops up allowing you to select the Components
and Groups to trace might not display a scroll bar, preventing you from
viewing all the components and groups.
The text area displaying the
selected components, groups, and trace levels does not have a vertical scroll
bar. This omission is a limitation of Netscape on a Solaris Operating Environment.
To
work around this problem, refresh the window to show the scroll bar.
Netscape screen blanks out while using the administrative
console
While working with the right-hand panel of the administrative
console to do regular administrative tasks, the browser screen blanks out
intermittently.
To work around this problem, do one of the following:
- After the problem occurs, close the Netscape browser, log in again, and
continue working.
- Use the Internet Explorer browser from a Windows machine.
- Use Netscape 7.x, Mozilla 1.x, Opera 5, or Konquerer browsers on the platform,
depending on which is available. Although there is not formal support for
these browsers, they have all been used successfully with the product and
in many cases work better than the previous 4.7.x series of Netscape browsers.
Resizing Netscape Version 4.7 causes errors
You
receive the following error messages when resizing Netscape Version 4.7:
Error 0
An error occurred while processing request: http://localhost:9090/admin/upload.do
message:
Details
com.ibm.webshpere.servlet.error.ServletErrorReport:
at java.lang.Class.newInstance0(Native Method)
... After resizing Netscape 4.7, Netscape has to reload
the page just as it initially loads the page on the first request. For pages
that do not expect POST data, it is not a problem. But for pages that do,
Netscape 4.7 cannot retain the data.
Solutions for using a heterogeneous
cell with iSeries nodes when running WebSphere Application Server Network
Deployment
The following information describes the problems and
solutions for using a heterogeneous cell with iSeries nodes when running WebSphere
Application Server Network Deployment:
When running WebSphere Application
Server Network Deployment on a non-iSeries system, for example, Windows NT,
AIX and Solaris Operating Environment, application servers or cluster members
created for a managed iSeries node do not start if you select to base the
application server or cluster member on a template. Application servers or
cluster members also do not start when you base the application server or
cluster member on an existing server, that is not part of an iSeries node.
Note:
If you base the new application server or cluster member on an application
server that already exists in the iSeries node, or on an application server
which is part of another iSeries node, the server starts and functions correctly.
To
work around this problem, after creating the application server or cluster
member, and before attempting to start the server, perform the following steps
from the WebSphere Application Server Network Deployment administrative console:
- Click Servers > Application Servers.
- Click the link for the server you just created.
- Click Process Definition.
- Change the value for the Executable Name field to /QSYS.LIB/QEJBAS5.LIB/QEJBSVR.PGM.
- Click Apply.
- Click Java Virtual Machine under the Additional Properties section.
- Change the value for the Initial Heap Size field to 96.
- Change the value for the Maximum Heap Size to 0.
- Click Apply.
- Click Application Servers > your server > Process Definition.
- Click Web Container.
- Click HTTP Transports under the Additional Properties section.
- Select the check box for the port that shows True in the SSL
Enabled column, and click Delete.
Note: You only need to do
step 13 if you do not have the Digital Certificate Manager product (5722SS1
option 34) and a cryptographic provider product (5722AC2 or 5722AC3) installed
on your iSeries server.
- Click Save link in the Messages box at the top of the workspace
frame.
- Select Synchronize with Nodes.
- Click Save.
- When running WebSphere Application Server Network Deployment on an iSeries
system, application servers or cluster members created for a managed non-iSeries
node, for example Windows NT, AIX and Solaris Operating Environment, do not
start if you base the application server or cluster member on a template.
Application servers or cluster member also do not start when you base the
application server or cluster member on an existing server that is part of
an iSeries node.
Note: If you base the new application server or cluster member
on an application server that already exists in the non-iSeries node, or on
an application server that is part of another non-iSeries node, the server
starts and functions correctly.
To work around this problem, after creating
the application server or cluster member, and before attempting to start the
server, complete the following steps from the WebSphere Application Server
Network Deployment administrative console:
- Click Servers > Application Servers.
- Click the link for the server you just created.
- Click Process Definition.
- Change the value for the Executable Name field to ${JAVA_HOME}/bin/java.
- Click Apply.
- Click Java Virtual Machine under the Additional Properties section.
- Change the value for the Initial Heap Size field to 0.
- Change the value for the Maximum Heap Size to 256.
- Click Apply.
- Click Application Servers > your server > Process Definition.
- Click Process Execution under the Additional Properties section.
- Change the Process Priority value to 20.
- Remove the value (QEJBSVR) specified for the Run As User field.
- Click Apply.
- Click Application Servers > your server > Process Definition.
- Click Web Container.
- Click HTTP Transports under the Additional Properties section
- Create a new HTTP transport that is Secure Sockets Layer (SSL) enabled.
Note:
You only need to do step 18 if you wish to use SSL with your Web server, or
you are using Single Sign On (SSO) support.
- Click Save link in the Messages box at the top of the workspace
frame.
- Select Synchronize with Nodes.
- Click Save.
Application Assembly Tool
All platforms
Help files are viewable only from a locally installed
browser
If you access any of the IBM WebSphere Application Server
tools from a remote machine, for example, the Application Assembly Tool, the
remote browser cannot display the help files. You can only view the help files
from a locally installed browser.
To work around this problem, close
all the Netscape sessions on the remote machine and click Help. A new
Netscape session starts and you can then view the Help files.
Information you should know about Application
Assembly Tool security roles
When you use the Application Assembly
Tool (AAT) at an application level (EAR file), security roles are synchronized
with the security roles defined for the embedded modules of the application.
If
a security role is manually added to the EAR file, you can automatically remove
it when the file is saved, if an embedded module does not reference the role
or the role is in conflict with an existing role. In this case, remove the
manually added role. All roles with the same name then are removed.
The
role is automatically added again when the file is saved, if it is still referenced
in an embedded module file. If a duplicate role is added in an embedded module
file, delete all roles with the same name and the correct role manually read
again.
Application Assembly Tool displays unsupported
type for application client resource reference
When configuring
a resource reference for an application client module in the Application Assembly
Tool, the Type field menu lists javax.resource.cci.ConnectionFactory
as an available resource reference. This type is not supported by the J2EE
application client run time. The supported types are:
- java.net.URL
- javax.mail.Session
- javax.jms.QueueConnectionFactory
- javax.jms.TopicConnectionFactory
- javax.jms.Queue
- javax.jms.Topic
- javax.sql.DataSource
Running the Application Assembly Tool on UNIX
platforms causes errors
A sample
of the errors that display follows:
...Font specified in font.properties not found [-urw-itc
zapfdingbats-medium-r-normal--*-%d-*-*-p-*-sun-fontspecific]
Font specified in font.properties not found [-urw-itc
zapfdingbats-medium-r-normal--*-%d-*-*-p-*-sun-fontspecific]
...
The Application Assembly Tool or installer functions are
not affected by these errors. These messages display in the command shell
that spawned the Java GUI. You can disregard these messages.
Apply interim fix PQ73432 to enable the configuration
of one-way relationships into read ahead hints
Read ahead hints
are defined in the WebSphere Application Server Enterprise Assembly Tool by
defining a path of container-managed relationships. The Version 5 and Version
5.0.1 of this tool incorrectly disallow the configuration of one-way relationships
for read ahead hints. If an application has defined one-way relationships,
these relationships will not be exposed in the list of available relationships
to configure into a read ahead hint.
This limitation is corrected in
WebSphere Application Server, Version 5.0. 2.
Apply interim fix PQ73432
to enable the configuration of one-way relationships into read ahead hints
in WebSphere Application Server, Version 5 and Version 5.0.1.
Databases-DB2
All Platforms
Illegal conversion occurs on any VARCHAR FOR
BIT DATA column in a container-managed persistent bean
When enterprise
beans with container-managed persistent (CMP) types that have any VARCHAR
FOR BIT DATA columns defined on a DB2 table are deployed in the DB2 universal
JDBC type 4 driver to persist the data, an SQLException of illegal conversion
is thrown at run time.
This exception only occurs when you use the DB2
universal JDBC type 4 driver and with the deferPrepares property
being set to true. When the deferPrepares property is set to true,
the DB2 universal JDBC type 4 driver uses the standard JDBC data mapping.
Currently,
the generated deployed code does not follow the standard JDBC specification
mapping. The failure at execution time is because of a problem in the tool
that prepared the enterprise beans for execution.
To avoid receiving
this exception, choose one of the following options:
- Set the deferPrepares property to false in the data
source configuration.
- Do not use the DB2 universal JDBC type 4 driver if your table has any
VARCHAR FOR BIT DATA or LONG VARCHAR FOR BIT DATA columns. Use the DB2 legacy
CLI-based JDBC driver to persist the data.
Refer to DB2 V8.1 readme for more details.
Incorrect user ID or password can result in
an error message
At the DB2 FP7 level, when configuring a 4.0 data
source for EJB 1.1 and 1.1 container-managed persistence (CMP) beans, and
the user ID or password is incorrect, you can get the following error message
in the SystemOut.log file while starting the application:
[7/22/02 17:37:13:547 CDT] 2827798b ConnectionPoo E CONM6009E: Failed to get connection to database from datasource (DB2PolicyDataSource).
[7/22/02 17:37:13:609 CDT] 2827798b JDBCPersister W CNTR0031W: Error starting CMP bean Life Insurance MD#lipolicy11.jar#IDTable:com.ibm.ejs.persistence.EJSPersistenceException: ; nested exception is:
COM.ibm.db2.jdbc.DB2Exception: [IBM][CLI Driver] SQL1403N The username and/or password supplied is incorrect. SQLSTATE=08004
COM.ibm.db2.jdbc.DB2Exception: [IBM][CLI Driver] SQL1403N The username and/or password supplied is incorrect. SQLSTATE=08004
To
work around this problem, verify the user ID or password and restart the application.
Mismatched user IDs causes a table not found or
object undefined exception from DB2
You can experience a
table not found or object undefined exception
from DB2, when the user ID that is used for the connection does not match
the user ID used to create a table.
If you use getConnection (user ID,
password) to get a connection and the user ID (for example, user ID2) is different
from the one (for example, user ID1) that is defined in the data source, then
user ID2 replaces user ID1 as the default in the data source. When your application
tries to get another connection without using the user ID and password, the
database uses user ID2 by default to get the connection. If you try to access
an object created by user ID1, the exception occurs.
To avoid receiving
the exception, do not use the user ID and password to get a connection. If
your application needs to connect this way, every getConnection() method should
always use the user ID and password to get a connection.
This problem
is fixed in DB2 V8.1 FP2
Using the administrative console to create a DB2
JDBC provider for a zSeries server causes an exception to occur
When
you use the administrative console to create a DB2 390 JDBC provider or a
DB2 390 JDBC provider (XA), an exception occurs with the message Name
is not a valid entry.
To work around this problem, choose the normal
DB2 JDBC provider to create a JDBC provider since they both use the same implementation
class. When you create a data source under this provider, change the data
store helper value from DB2DataStoreHelper to DB2390DataStoreHelper. Otherwise,
your database access does not work.
DB2 Universal JDBC type 4 driver is supported
DB2
Universal JDBC type 4 driver is supported in WebSphere Application Server,
Version 5.0.1. The minimum fix pack level is DB2 V8.1 FP1. This driver does
not support XA transaction. It is only supported in WebSphere Application
Server, Version 5.0.1 data source.
There is a known problem on mapping StaleConnectionException.
When you see the following exception stack, it is mapped to StaleConnectionException:
Error Code = 0
SQL State = null
com.ibm.db2.jcc.b.DisconnectException: A communication error has been detected. Communication protocol being used: {0}. Communication API being used: {1}. Location where the error was detected: {2}. Communication function detecting the error: {3}. Protocol specific error codes(s) {4}, {5}, {6}. TCP/IP SOCKETS Agent.sendRequest() OutputStream.flush() Connection reset by peer: socket write error * 0
at com.ibm.db2.jcc.c.a.a(a.java:329)
at com.ibm.db2.jcc.c.a.u(a.java:302)
The problem has been fixed in DB2 V8.1 FP2.
Accessing
two different rows in a table might cause deadlock
Deadlock results
from accessing two different rows in a DB2 table when the next key is not
locked for all INSERT and DELETE statements.
To eliminate the deadlock,
set the DB2_RR_TO_RS environment variable from the DB2 command line window.
Setting this environment variable causes the following:
- If RepeatableRead (RR) is your chosen isolation level, it is essentially
downgraded to Read Stability (RS).
- If you chose a different isolation level and the DB2_RR_TO_RS environment
variable is turned on, scans of the database skip rows that are deleted but
not committed. This activity occurs even if the row qualified for the scan.
The skipping behavior affects the RR, RS, and Cursor Stability (CS) isolation
levels.
Possible scenario
Transaction A deletes the row
with column1=10 and transaction B does a scan
on column1>8 and column1<12.
With DB2_RR_TO_RS=OFF, transaction B waits for
transaction A to commit or roll back. If transaction A rolls back, the row
with column1=10 is included in the result set
of the transaction B query. With DB2_RR_TO_RS=ON transaction B does
not wait for transaction A to commit or roll back. It immediately receives
query results that do not include the deleted row.
As a workaround,
set the DB2_RR_TO_RS enviroment variable to ON by typing DB2_RR_TO_RS=ON on
the command line.
Information about the StaleConnectionException
on Linux systems
The StaleConnectionException
SQl1224 is related to the extension shared memory attachment. Linux systems
have a semaphore problem causing the DB2 SQL1224 error.
To work around
the problem, set the loopback for your database. For example, if your database
is WAS, host name is LHOST, database service port number is 50000, issue the
following commands from the DB2 command-line window:
db2 catalog TCPIP node RHOST remote LHOST server 50000
db2 catalog db WAS as WASAlias
db2 uncatalog db WAS
db2 catalog db WASAlias as WAS at node RHOST
Verify this information by issuing the following commands from
the DB2 command-line window:
db2 connect to WAS user xxx
passwd: xxx
where xxx is the user
name or password
Setting connection autoCommit to false does
not work correctly in a global transaction or a local transaction when you
use the DB2 Universal JDBC provider driver
Setting connection autoCommit
to false does not work correctly in a global transaction or a local transaction
when you use the DB2 Universal JDBC provider driver.
In a global transaction:
When
you call a connection.setAutoCommit(false) in a global transaction, an incorrect
exception DSRA9350E: Operation setAutoCommit is not allowed during a
global transaction is thrown.
In a local transaction:
When
you call cconnection.setAutoCommit(false), you can do work to the connection
and call connection.setAutoCommit(false) again on the same connection. The
second connection.setAutoCommit(false) commits the previous work. This behavior
is the same as the behavior of the native DB2 universal JDBC driver but is
not consistent with the behaviors when you use other JDBC drivers in WebSphere
Application Server. The correct WebSphere Application Server behavior is
that the second connection.setAutoCommit(false) does not commit the previous work.
Apply
interim fix PQ73523 to solve this problem.
Databases- IBM Cloudscape
Application table lock can occur in IBM Cloudscape
when using RepeatableRead
When your application uses a new Version
5 data source on the IBM Cloudscape database, your application can acquire
a table lock when the application tries to use a RepeatableRead isolation
level to select a row in a table for update.
In IBM Cloudscape Version
5.0, the "Select..ForUpdate" statement with a RepeatableRead isolation level
locks the entire table. Another transaction is blocked when it tries to access
the same table. The transaction remains blocked until either the transaction
holding the table lock releases the lock by committing or aborting the transaction,
or when the lock wait times out. If the lock wait results in a deadlock condition,
one transaction aborts and the other transaction continues.
To work
around this problem, take the following action:
- Beans using bean-managed persistence (BMP) or JDBC application:
Define
an index or a key and change the select query to have the "of column" clause,
where the columns are not key fields in the index or key. For example:
create table foo (a int, b int);
create index foo_index on foo (a);
// create a cursor using of column clause to get a row lock
select * from foo where a=100 for update of b;
- Beans using container-managed persistence (CMP):
Avoid using the access
intent of wsPessimisticUpdate profile. This profile results in a RepeatableRead
isolation level on a connection.
The locale JAR files for IBM Cloudscape are available
IBM
WebSphere Application Server now ships the locale JAR files for IBM Cloudscape.
The files are located in the <install_root>lib/cloudscape directory.
To use a specific language, add the respective locale JAR file to the class
path in both the system class path and the administrative console. The class
path of the system is needed for the IBM Cloudscape GUI, for example, cview.
The administrative console is needed for the error messages thrown when running
IBM WebSphere Application Server.
Running IBM WebSphere Application Server with
IBM Cloudscape causes errors
When running IBM WebSphere Application
Server with IBM Cloudscape, you can see the following exception:
SQLException Info: ErrorCode=20000, State=XJ055,
Message: This connection is not attached to any transaction, make sure it has not been ended by the XAResource.
To
fix this problem, verify the statement cache size is 0 on the data
source. The default value is 0.
IBM Cloudscape is not licensed for use in a production
environment
IBM WebSphere Application Server provides a copy of
the IBM Cloudscape database for development use. IBM Cloudscape is a small
footprint database written entirely in the Java language for platform independence.
Developers can use IBM Cloudscape to test Web applications that need to access
data.
IBM Cloudscape is not licensed for use in a production environment.
For
information about IBM Cloudscape, visit the IBM Cloudscape Web site at: http://www.ibm.com/software/data/cloudscape/pubs/collateral.html
Non-English locale of cview does not display
the information tab
When you use cview in a non-English locale,
IBM Cloudscape has a problem displaying the contents of the information tab.
To
work around this problem, view the system information by issuing the command
line version. The command-line version, called sysinfo.bat or sysinfo.ksh, ships with IBM WebSphere Application
Server. If you plan to use the command-line, set the respective locale JAR
file in your class path.
Set JAR file names on class path to use cview
During
IBM WebSphere Application Server installation, the setEmbeddedCP.bat file
is not installed, therefore you cannot use cview.bat or cview.ksh to set the JAR files to the class path.
In order to use cview.bat or cview.ksh,
you need to set the following JAR files on the class path:
<install_root>/lib/db2j.jar
<install_root>/lib/db2jcview.jar
<install_root>/lib/db2jtools.jarSee the setEmbeddedCP.bat file for an example of how to set
these files.
When IBM Cloudscape rolls back a transaction,
the IBM WebSphere Application Server might display an XANOTA exception
When
IBM Cloudscape rolls back a transaction, it might throw an XANOTA exception,
which the IBM WebSphere Application Server displays.
You can ignore
this exception.
Databases-Informix
An XA exception is thrown during an XA recovery
This
section describes the restriction and solution for a multi-bit flag sent to
an XA recovery.
An XA exception is thrown during an XA recovery due
to Informix defect 161882. This exception can be caused by one of the following
conditions:
- A shutdown of WebSphere Application Server with an application still running
XA transactions.
- A WebSphere Application Server crashes.
An example of the exception is as follows:
33433c9f XARminst F WTRN0017W: The transaction service encountered an error on an xa_recover operation. The resource was CMXAResourceInfo: DataSource Properties [com.informix.jdbcx.IfxXADataSource]: {user=informix;password=XXXXXXXX;description=DataSource for LifeInsApp;portNumber=1527;databaseName=ann;ifxIFXHOST=ifxHost;serverName=ol_ifxHost;}. The error code was XAER_INVAL.
The exception stack trace
follows:
javax.transaction.xa.XAException
at com.informix.jdbcx.IfxXAResource.recover(IfxXAResource.java:310)
at com.ibm.ejs.jts.jta.XARminst.recover(XARminst.java:126)To
avoid this error, delete the WebSphere Application Server transaction log
in the /<WAS_HOME>/tranlog directory before bringing the WebSphere
Application Server on-line.
Informix JDBC driver incorrectly handles escaped
newline character
Informix JDBC driver incorrectly handles inserting
quoted ANSI control characters into column types CHAR and LVARCHAR. The control
characters stored into the column might not be correctly retrieved.
Informix JDBC driver incorrectly handles ANSI
control characters in a SELECT statement
Informix JDBC driver incorrectly
handles ANSI control characters in a SELECT statement. A statement like "SELECT
* FROM TABLE1 WHERE TESTCOL = '\n'''" produces the following SQLException:
SQL State: IX000
Error Code: -282
java.sql.SQLException: Found a quote for which there is no matching quote.
The Informix JDBC provider produces errors when
writing character data to a Character Large Object file
When attempting
to write character data to a Character Large Object (CLOB) field using the
setCharacterStream method on the Informix JDBC provider, the following exception
occurs:
java.sql.SQLException: Can't convert fromnull
at com.informix.util.IfxErrMsg.getSQLMinorException(IfxErrMsg.java:511)
at com.informix.jdbc.IfxObject.fromCharacterStream(IfxObject.java:856)
at com.informix.jdbc.IfxValue.makeInstance(IfxValue.java:1091)
at com.informix.jdbc.IfxPreparedStatement.setCharacterStream(IfxPreparedStatement.java:3202)
at JDBCDriverTest.main(JDBCDriverTest.java:69)This problem
has been reported to Informix.
Exception thrown when running an XA transaction
When
running an XA transaction involving two connections to the same database,
the following pattern results in an exception:
get XA connection 1
get XA connection 2
start XA resource for connection 1
start XA resource for connection 2
Connection 1: insert a new record into a table T1
Connection 2: update an existing record in the same table --> exception is thrown here
You
get the following exception:
java.sql.SQLException: ISAM error: record is locked.
java.sql.SQLException: Could not do a physical-order read to fetch next row.
at com.informix.jdbc.IfxSqli.addException(IfxSqli.java:3021)
at com.informix.jdbc.IfxSqli.receiveError(IfxSqli.java:3335)
at com.informix.jdbc.IfxSqli.dispatchMsg(IfxSqli.java:2288)
at com.informix.jdbcx.IfxXASqli.receiveMessage(IfxXASqli.java:119)
at com.informix.jdbc.IfxSqli.executeCommand(IfxSqli.java:720)
at com.informix.jdbc.IfxResultSet.executeUpdate(IfxResultSet.java:317)
at com.informix.jdbc.IfxStatement.executeUpdateImpl(IfxStatement.java:838)
at com.informix.jdbc.IfxPreparedStatement.executeUpdate(IfxPreparedStatement.java:269)
at InfXATest.run2(InfXATest.java:289)
at InfXATest.main(InfXATest.java:11)Currently, there
is no work around for this problem. The problem has been reported to Informix.
OUT parameters not supported in Informix
Dynamic Server 9.3
The Informix Dynamic Server 9.3 does not support
multiple OUT parameters in functions or procedures as required by the JDBC
2.0.
PTS#154442 has been opened with Informix to resolve this problem.
LOCATE function not supported in Informix
Dynamic Server 9.3
The Informix Dynamic Server 9.3 does not support
the locate function as required by the JDBC 2.0.
Two-phase commit transactions cause Informix
Dynamic Server 9.3 to crash
Two-phase commit transactions sometimes
cause Informix Dynamic Server 9.3 to crash. This problem is addressed in Informix
bugs 158717 and 158860.
Prepared statement cache does not work
with two-phase commit and Informix Dynamic Server (IDS) 9.3 data sources
Prepared
statement caching does not work with two phase commit and Informix Dynamic
Server 9.3 data sources. This will be addressed in a future release of IDS.
INOUT parameters not supported in Informix
Informix
Dynamic Server does not support INOUT parameters in functions or procedures
as required by the JDBC 2.0.
PTS#159522 has been opened with Informix
to resolve this problem.
Binary Large Object not supported in
Informix
Informix Dynamic Server does not support Binary Large Objects
(BLOBS) in the DISTINCT clause of a SELECT statement as required by the JDBC
2.0.
PTS#159333 has been opened with Informix to resolve this issue.
Databases-Microsoft SQL Server
Microsoft JDBC Driver for SQL Server 2000
All
platforms
Use the proper DataStoreHelper class to configure
data sources for the Connect JDBC driver
If you use the WebSphere
Embedded Connect JDBC driver for Microsoft SQL Server, you must use the proper
DataStoreHelper class. If you use an incorrect DataStoreHelper class, either
of the following exceptions might occur:
java.sql.SQLException: [IBM][SQLServer JDBC Driver]This driver is locked for use with embedded applications.
at com.ibm.websphere.jdbc.base.BaseExceptions.createException(Unknown Source)
at com.ibm.websphere.jdbc.base.BaseExceptions.getException(Unknown Source)
at com.ibm.websphere.jdbc.base.BaseConnection.validatedUnlocked(Unknown Source)
java.sql.SQLException: [IBM][SQLServer JDBC Driver]?}???h???C?o???A?????????A?v???P?[?V?????????g?p?????b?N???????E?B
at com.ibm.websphere.jdbc.base.BaseExceptions.createException(Unknown Source)
at com.ibm.websphere.jdbc.base.BaseExceptions.getException(Unknown Source)
at com.ibm.websphere.jdbc.base.BaseConnection.validatedUnlocked(Unknown Source)
The correct DataStoreHelper class for the WebSphere Embedded Connect
JDBC driver for Microsoft SQL Server and WebSphere Embedded Connect JDBC driver
for Microsoft SQL Server (XA) providers is com.ibm.websphere.rsadapter.WSConnectJDBCDataStoreHelper.
MDAC 2.6 SP1 is a prerequisite for SequeLink Server
5.3
MDAC 2.6 SP1 is a prerequisite for SequeLink Server 5.3. You
can download this prerequisite from the Microsoft Web site at http://www.microsoft.com
.
Microsoft SQL Server 7 error occurs when using
UPDLOCK hint
When running SequeLink 5.3 against SQLServer 7 and
setting ResultSet concurrency = CONCUR_UPDATABLE, ResultSet type
= TYPE_SCROLL_SENSITIVE, the following error occurs when executing a statement
containing the UPDLOCK locking hint:
SQL State: 42000
Error Code: 16940
java.sql.SQLException: [IBM][SequeLink JDBC Driver][SQL Server]Cannot specify UPDLOCK or TABLOCKX with READ ONLY or INSENSITIVE cursors.
This error is caused by a bug in Microsoft SQL Server 7. The SequeLink
5.3 driver provides the following workaround for this problem:
- Open the Microsoft Management Console for SequeLink 5.3 ( %SL53%\admin\sladmin53.msc ).
- Expand SLAgent53 > SequeLink Services > SLSQLServer53 > Configuration > Data
Source Settings > Default.
- Right-click Default and then click New > Attribute.
- Click DataSourceMSSCursortype and fill in a value of Clientside.
- Click OK to add the attribute and then select the yellow check
mark box at the top of the console to save the changes.
- The property appears in the Advanced folder under the Default
data source, if you want to change or delete it later.
Instructions for installing the SequeLink 5.3
server
The WebSphere Application Server, Version 5 InfoCenter does
not include documentation for installing the SequeLink 5.3 server that ships
with WebSphere Application Server. Visit the following Web page of DataDirect
Technologies driver for instructions on installing the SequeLink 5.3 server:
ftp://ftp.software.ibm.com/software/websphere/info/tools/DataDirect/datadirect.htm
The DataDirect Technologies SequeLink 5.3 JDBC
driver can ignore local transactions
The DataDirect Technologies
SequeLink 5.3 JDBC driver contains a bug which can ignore local transaction
rollbacks after a connection is used in a global transaction. DataDirect
Technologies has provided a patch for this problem, which you can download
from:
ftp://ftp.software.ibm.com/software/websphere/info/tools/DataDirect/datadirect.htm
Do not use DBUser and DBPassword
Do
not use DBUser and DBPassword or HUser and HPassword data source properties
for SequeLink 5.3. User names and passwords specified this way are exposed
in WebSphere traces and XML files. Instead, specify this information by creating
an alias for the data source.
Additionally, even if you are not concerned
about exposed passwords and DBUser and DBPassword are specified, SequeLink
does not honor the values used, so this method of specifying user names and
passwords is unreliable.
The DataDirect Technologies SequeLink JDBC
driver hangs when running in two-phase commit mode
The DataDirect
Technologies SequeLink JDBC driver can hang when it runs in two-phase commit
mode against Microsoft SQL Server 7. Case 15005170 is open with DataDirect
Technologies for this problem.
To work around this problem, take one
of the following actions:
- Using the DataDirect Technologies Connect JDBC driver connect to the Microsoft
SQL Server 7 instead of the SequeLink.
- Upgrade to Microsoft SQL Server 2000 and continue using the SequeLink
JDBC driver to connect.
- Use a SequeLink data source enabled for one-phase commit mode and recode
your application to use only one-phase commit mode.
- When a patch becomes available from DataDirect Technologies, download
a corresponding patch for the WebSphere Application Server embedded SequeLink
driver from the DataDirect Technologies driver update download page. The
URL for this page is: ftp://ftp.software.ibm.com/software/websphere/info/tools/DataDirect/datadirect.htm
The Microsoft JDBC driver for Microsoft
SQL Server 2000 fails
The Microsoft JDBC driver for Microsoft SQL
Server 2000 fails with a java.lang.NoClassDefFoundError when DataSource.setLogWriter is used, spyAttributes are
specified, or IBM WebSphere Application Server tracing is enabled.
The
problem occurs because Microsoft does not include the spy.jar file
in its JDBC driver for Microsoft SQL Server 2000 despite the fact that the
driver requires classes found in the file spy.jar to
function properly. Consequently, the Microsoft driver fails with the NoClassDefFoundError when
IBM WebSphere Application Server tracing is enabled. Workaround options follow:
- Open a problem report with Microsoft.
- Use the embedded WebSphere ConnectJDBC driver instead of the Microsoft
driver.
- Obtain a compatible spy.jar file from some
other source. Neither of the spy drivers which ship with the IBM WebSphere
Application Server (intended only for use with the WebSphere embedded drivers)
are compatible with the Microsoft driver.
- Avoid all uses of DataSource.setLogWriter, spyAttributes, and IBM WebSphere Application Server
tracing.
The Microsoft driver for Microsoft SQL Server
2000 ignores setAutoCommit(true) during a local
transaction
The Microsoft driver for Microsoft SQL Server 2000 does
not honor setAutoCommit(true) during a local transaction.
The driver continues to operate as though autoCommit is false while reporting
a value of autoCommit = true. This behavior can cause unexpected activity
since it is inconsistent with the JDBC specification. According to the JDBC
specification, setting autoCommit to true during a local transaction
implicitly commits any outstanding local transaction and changes the connection
to autoCommit enabled (true) mode.
Perform one of the following
steps to work around the problem:
- Open a problem report with Microsoft.
- Use the embedded WebSphere ConnectJDBC driver instead of the Microsoft
driver.
- Do not rely on the use of setAutoCommit(true) to
end local transactions.
Databases-Oracle
All platforms
Oracle JDBC driver memory leak
If you
use Oracle 9i R2, you can encounter memory leaks when using the Oracle XA
JDBC driver.
To fix this problem, visit the Oracle Metalink Web site
and download Oracle Patchset 92020.
Oracle rollbacks fail with bean-managed persistence
and unshareable mode
When running a Global transaction with an unshareable
mode where two or more connections are created as part of the Global transaction,
Oracle throws the exception "ora-02051: another session
in same transaction failed" when doing a rollback or commit on the Global
transaction.
Use Sharable mode when running a bean-managed persistence
(BMP) or servlet to work around the problem as follows:
- BMP:
Set the res-sharing-scope tag to Sharable in the ejb-jar.xml file,
which is found in the META-INF directory of the yourbean.jar file. For example:
<res-sharing-scope>Shareable</res-sharing-scope>
- Servlet:
Set the res-sharing-scope tag to Sharable in the web.xml file,
which is found in the WEB-INF directory of the .war file. For example:
<res-sharing-scope>Shareable</res-sharing-scope>
This situation is a recognized Oracle problem. The Oracle bug
number is 2511780.
Error message displays when inserting or updating
a LOB data type
When using the LOB columns (BLOBS and CLOBS) with
Oracle 8i, you cannot use setCharacterStream or setBinaryStream to set BLOBS
and CLOBS when the size of the LOB columns is greater than 4k.
You receive
one of the following error messages when inserting or updating a LOB (BLOBS
and CLOBS) that are greater than 4k, depending on the Oracle release:
"Data size bigger than Max size for this type: xxxx. SQL State = <null>, Error Code = 17,070"
"java.sql.SQLException: Io exception: End of TNS data channel SQL State: null Error Code: 17002"
Oracle
has lifted the restriction in the 9i R2 OCI driver (thick driver). If you
have one of these error messages or you know that your data exceeds 4k, update
to Oracle 9i R2 and use the OCI driver when connecting to Oracle server. If
you cannot update to Oracle 9i R2, contact Oracle Support.
Setting the correct value for autoCommit in
the connection of the OracleXADataSource
When your application tries
to use the OracleXADataSource to run in an autoCommit enabled mode, set the
autoCommit value to TRUE once a connection is obtained from the OracleXADatasource.
The default autoCommit value for every connection from the XA data source
is FALSE. This value is different from the OracleConnectionPoolDataSource,
which has the default autoCommit value of TRUE.
If your application
runs in a global transaction, set the autoCommit value to FALSE to
start a global transaction.
Bug#2618770 has been accepted by Oracle
for this problem.
Using OCI drivers requires 32-bit Oracle libraries
If
you are using the OCI drivers, you must use 32-bit Oracle libraries.
Set
the OCI driver on Oracle as follows:
- (Optional, might already exist) Update entries in the listener.ora file
on the server side. You can find the DESCRIPTION_LIST and SID_LIST in the $ORACLE_HOME/network/admin directory.
Modify the entries in the << >>.
Add the following to the DESCRIPTION_LIST entry:
(DESCRIPTION =
(ADDRESS = (PROTOCOL = TCP) (HOST = <<host_name>>) (PORT = 1521))
)
Add the following to the SID_LIST:
(SID_DESC =
(GLOBAL_DBNAME = <<database_name >>)
(ORACLE_HOME = <<oracle_home>> )
(SID_NAME = <<database_name>>)
)Note: An example of <<database_name>> is "was".
An example of <<oracle_home>> is E:\Oracle\Ora81.
- Update entries in the tnsnames.ora file located
in $ORACLE_HOME/network/admin directory. You are
modifying the entries in the << >>.
Add the following to the tnsnames.ora file:
tnsname =
(DESCRIPTION =
(ADDRESS_LIST =
(ADDRESS = (PROTOCOL = TCP) (HOST = <<server_name>>) (PORT = 1521) )
)
(CONNECT_DATA =
(SERVICE_NAME = <<database_name >>)
)
)Note: An example of <<database_name>> is "was".
- Add the location where Oracle is installed in the .profile export,
for example ORACLE_HOME:<oracle_root_install>.
- Add the following to the Native Library Path in the administrative console:
For
AIX systems:
export LD_LIBRARY_PATH=$ORACLE_HOME/lib32
For
Solaris Operating Environment and other platforms:
export PATH=$ORACLE_HOME/lib32
For
Hewlett Packard environment:
export SHLIB_PATH=$ORACLE_HOME/lib32
- Add the following to the class path of the administrative console:
export CLASSPATH=$ORACLE_HOME/jdbc/lib/classes12.zip: $ORACLE_HOME/jdbc/lib/nls_charset12.zip
- Set the URL inside the administrative console using the following format:
jdbc:oracle:oci8:@<tnsname>
where <tnsname> is
what is defined in the tnsames.ora file
Oracle XA driver displays an error when using an
autoCommit true value followed by a global transaction
When running
the IBM WebSphere Application Server against an Oracle driver, problems occur
when the Oracle XA driver uses autoCommit with a value of true, followed
by a global transaction.
If you encounter this problem, set autoCommit
to false before starting the global transaction on the same connection.
TAR
2603121.995 has been opened with Oracle to resolve this problem.
Using PreparedStatement.setObject fails for some
types with ClassCastException
Using PreparedStatement.setObject
fails for some types with ClassCastException when using Oracle.
The
following incorrect uses of setObject fail with ClassCastException instead
of successfully updating the table:
// Test use the following tables
create table Bit_Tab (MAX_VAL SMALLINT, MIN_VAL SMALLINT, NULL_VAL SMALLINT NULL);
create table Decimal_Tab (MAX_VAL DECIMAL(30,15),MIN_VAL DECIMAL(30,15), NULL_VAL DECIMAL(30,15) NULL);
create table Bigint_Tab (MAX_VAL NUMBER(19,0), MIN_VAL NUMBER(19,0), NULL_VAL NUMBER(19,0) NULL);
create table Timestamp_Tab (IN_TIME DATE, NULL_VAL DATE NULL);
// Tests use the following prepared statements
pstmt1i = con.prepareStatement("insert into Bit_Tab VALUES(?, ?, ?)");
pstmt2i = con.prepareStatement("insert into Decimal_Tab VALUES(?, ?, ?)");
pstmt3i = con.prepareStatement("insert into Bigint_Tab VALUES(?, ?, ?)");
pstmt4i = con.prepareStatement("insert into Timestamp_Tab VALUES(?, ?)");
pstmt5i = con.prepareStatement("insert into Decimal_Tab VALUES(?, ?, ?)");
// failing code snippets
// Test 1 - Boolean
pstmt1i.setBoolean(1,true);
pstmt1i.setBoolean(2,true);
java.math.BigDecimal decimalVal = new java.math.BigDecimal(1);
Boolean minDecimalVal=new Boolean(decimalVal.toString());
pstmt1i.setObject(3,minDecimalVal,java.sql.Types.DECIMAL,2);
pstmt1i.executeUpdate();
// Test 2 - Float
pstmt2i.setFloat(1,2.1F);
pstmt2i.setFloat(2,2.2F);
String stringVal = "1.1";
Float floatVal = new Float(stringVal);
pstmt2i.setObject(3,floatVal,java.sql.Types.DECIMAL,15);
pstmt2i.executeUpdate();
// Test 3 - Long
pstmt3i.setFloat(1,3.1F);
pstmt3i.setFloat(2,3.2F);
Long val = new Long(1);;
String stringValLing = new String(val.toString());
Long longVal = new Long(stringValLing);
pstmt3i.setObject(3,longVal,java.sql.Types.CHAR);
pstmt3i.executeUpdate();
// Test 4 - Timestamp
pstmt4i.setTimestamp(1, new Timestamp(System.currentTimeMillis()));
pstmt4i.setTimestamp(2, new Timestamp(System.currentTimeMillis()));
Timestamp getTSval = new Timestamp(System.currentTimeMillis());
String stringValTs = getTSval.toString();
Timestamp timeStampVal = Timestamp.valueOf(stringValTs);
pstmt4i.setObject(2,timeStampVal,java.sql.Types.DATE);
pstmt4i.executeUpdate();
// Test 5 - BigDecimal
pstmt5i.setFloat(1, 5.1F);
pstmt5i.setFloat(2, 5.2F);
String stringValBd = "1.1";
stringValBd = stringValBd.trim();
java.math.BigDecimal bigDecimalVal = new BigDecimal(stringValBd);
pstmt5i.setObject(3,bigDecimalVal,java.sql.Types.CHAR);
pstmt5i.executeUpdate();To avoid getting the exception, change
the code to avoid using setObject.
This problem is being addressed by
Oracle in the bug report 2640294.
Oracle fails to throw SQLExceptions
Oracle
fails to throw an SQLException when the incorrect type of SQL statement is
used on either the PreparedStatement.executeUpdate() or PreparedStatement.executeQuery()
methods.
Only use SQL statements returning a count with the executeUpdate()
method, and only use SQL statements returning a single result set with the
executeQuery() method.
The database generates an SQLException if an
SQL statement fails to meet the requirement it passes to either executeUpdate
or executeQuery. When using Oracle, executeUpdate and executeQuery return
successfully when passed by an SQL statement that fails to meet the requirement.
When
running with Oracle, the following code snippet does not return SQLException:
pstmt1 = con.prepareStatement("select * from Bit_Tab");
pstmt2 = con.prepareStatement("insert into Bit_Tab VALUES(?, ?, ?)");
//Test 1 - executeUpdate using select
try
{
pstmt1.executeUpdate();
}
catch (SQLException e)
{
System.out.println("SQLException occurred");
}
// Test 2 - executeQuery using insert
try
{
pstmt2.setBoolean(1,true);
pstmt2.setBoolean(2,true);
pstmt2.setBoolean(3,true);
ResultSet rs = pstmt2.executeQuery();
}
catch (SQLException e)
{
System.out.println("SQLException occurred");
}TAR 2688461.999 has been opened with Oracle to resolve this
problem.
CONM6021W messages appear when an Oracle
9.2.0.2 server and a 2511780 patch are used
If you use an Oracle
9.2.0.2 server, apply the 2511780 patch, and create the transactionBranchesLooselyCoupled
custom property for a 4.0 data source, the CONM6021W messages appear
every time a 4.0 data source connection is created.
The CONM6021W message
is similar to the following example:
CONM6021W: An Oracle JDBC provider property (TransactionLooselyCoupled) has been set. Oracle patch 2511780 must be applied before setting this property.
You can see this message frequently in the server SystemOut.log file.
Oracle does not support the three-argument version
of the locate function in the Solaris Operating Environment J2EE compliance
test suite
While running
the IBM WebSphere Application Server in the Solaris Operating Environment
J2EE Compliance test suite against Oracle, the three-argument version of the
locate function is not supported by Oracle. The two-argument version is supported
by Oracle. According to Oracle, the three-argument version of the locate function
is not part of the J2EE specifications.
Oracle thin driver does not support null user
ID and password and this causes an SQLException
When you run WebSphere
Application Server with Oracle, if you do not set the J2C authentication alias
or you do not use getConnection (user ID and, password ), a null is passed
to Oracle for user ID and password. Since Oracle thin does not support null
user ID and password, Oracle throws the following exception:
java.sql.SQLException: invalid arguments in call
To
work around the problem, set the J2C authentication alias on the data source
and restart the server.
Databases-Sybase
Executing the DatabaseMetaData.getBestRowIdentifier()
method in an XA transaction causes errors
Executing the DatabaseMetaData.getBestRowIdentifier()
method while in an XA transaction causes the following errors:
SQL Exception: The 'CREATE TABLE' command is not allowed within a
multi-statement transaction in the 'tempdb' database. Calling
DatabaseMetaData.getBestRowIdentifier()
Currently, this method
fails when using Sybase. This problem occurs with other methods as well, including:
- getBestRowIdentifier();
- getVersionColumns();
- getTablePrivileges();
- getProcedureColumns();
- getPrimaryKeys();
- getIndexInfo();
- getImportedKeys();
- getExportedKeys();
- getCrossReference();
- getColumns();
- getColumnPrivileges();
Case 10880427 has been opened with Sybase to resolve this problem.
Sybase requirements for using the escapes and DatabaseMetaData
methods
To use the escapes and DatabaseMetaData methods, you must
install stored procedures on the Adaptive Server Enterprise or Adaptive Server
Anywhere database where you want to use these methods. These stored procedures
are also required by some of the connection methods.
To check for the
presence of LOCATE ():
- Open a Sybase isql command prompt.
- Type the command use master.
- Type the command go.
- Type the SQL command and select * from jdbc_function_escapes.
- Type the command go.
The following appears:
escape_name
map_string
----------------------------------------
----------------------------------------
abs
abs(%1)
acos
acos(%1)
asin
asin(%1)
atan
atan(%1)
atan2
atn2(%1, %2)
ceiling
ceiling(%1)
:::::::::::::::::::
:::::::::::::::::::
locate
charindex ((convert (varchar, %1)), (convert (varchar, %2)))If
the function does not exist, upgrade jConnect to at least Version 5.2 EBF
10635 and run the following command:
java IsqlApp -U sa -P -S jdbc:sybase:Tds:<hostname>:4100 -I %JDBC_HOME%\sp\sql_server12.sql -c go
Database deadlocks and XA_PROTO errors
occur when using Sybase
When using Sybase with the IBM WebSphere
Application Server, do one of the following to prevent database deadlocks
and errors:
- Change the transaction isolation level on the connection to TRANSACTION_READ_COMMITTED.
Set the isolation level on the connection for unshareable connections or,
for shareable connections, define the isolation levels in the resource reference
for your data source using the Application Assembly Tool (AAT).
- Modify Sybase by doing one of the following:
Executing a stored procedure containing a SELECT
INTO command causes exception
An attempt to execute a stored
procedure containing a SELECT INTO command results in the following
exception:
SVR-ERROR: SQL Exception SELECT INTO command not allowed within multi-statement transaction
Case
10868947 has been opened with Sybase to resolve this problem.
Error is incorrectly reported about IMAGE
to VARBINARY conversion
The following error is incorrectly reported:
com.sybase.jdbc2.jdbc.SybSQLException: Implicit conversion from data type 'IMAGE' to 'VARBINARY' is not allowed.
Use the CONVERT function to run this query.
The error is about
a VARBINARY column only and causes confusion if you also have an IMAGE column.
Do
one of the following to work around this problem:
- Use a PreparedStatement.setBytes() method instead of a PreparedStatement.setBinaryStream()
method
- Use a LONG VARBINARY for the column type if you want to continue using
the setBinaryStream() method. You might want to make this change because the
size limit for VARBINARY is 255 bytes.
For example:
// ***************CORRECTION*****************************
// setBinaryStream fails for column type of VARBINARY , use setBytes() instead
//stmt4.setBinaryStream(8,new java.io.ByteArrayInputStream(tempbyteArray),tempbyteArray.length);
stmt4.setBytes(8,tempbyteArray);Java Database Connectivity 1.0 standard methods
are not implemented and generate a SQL exception when used
The following
Java Database Connectivity (JDBC) 1.0 standard methods are not implemented
and generate a SQL exception when used:
- ResultSetMetaData.getSchemaName()
- ResultSetMetaData.getTableName() (implemented only for text and image
datatypes)
- ResultSetMetaData.getCatalogName()
Sybase transaction manager fails after trying
to alleviate a deadlock error
If an application encounters a deadlock,
Sybase detects the deadlock and throws an exception. Because of this detection,
the transaction manager calls an xa_end with a TMFAIL in it.
The call succeeds, but causes
another Sybase exception, XAERR_PROTO. This exception
only appears in the error log and does not cause any functional problems.
All applications should continue to run, therefore no workaround is necessary.
Case
10869169 has been opened with Sybase to resolve this problem.
Starting an XA transaction when the autoCommit
value of the connection is false causes error
The exception
thrown is javax.transaction.xa.XAException with
stack trace similar to the following:
at com.sybase.jdbc2.jdbc.SybXAResource.sendRPC(SybXAResource.java:711)
at com.sybase.jdbc2.jdbc.SybXAResource.sendRPC(SybXAResource.java:602)
at com.sybase.jdbc2.jdbc.SybXAResource.start(SybXAResource.java:312)This
problem affects you when you do both local and global transactions. If, in
a local transaction, the autoCommit default value is set to false,
and a global or XA transaction starts (either a user transaction started by
you, or a container transaction started by a container), the exception occurs.
This
problem is a Sybase bug as the start() method can fail unexpectedly, regardless
of the value of autoCommit. Currently, there is no workaround for this problem,
therefore it is not recommended that you mix local and global transactions.
Case 10880792 has been opened to resolve this problem.
Sybase does not throw an exception when an incorrect
database name is specified
Verify that your database name is correctly
entered on the data source properties.
Most databases (DB2, Oracle,
Informix , MS SQL Server and Cloudscape) throw an exception when the database
specified does not exist. But Sybase does not throw an exception when an incorrect
database name is specified. Sybase generates an SQL warning and then connects
to the default database. If you misspell the requested database name, Sybase
connects you to the master or the default database where the table you requested
is not found.
Interoperability
Interoperability of the handle
formats in WebSphere Application Server, Version 5 and Version 5.0.1
Applications
that attempt to persist handles to enterprise beans and EJBHome needed
to subclass ObjectInputStream in WebSphere Application Server, Version 5.
This action was required so that the subclass ObjectInputStream could utilize
the context class loader to resolve the classes for enterprise beans and EJBHome
stubs.
In addition, handles created and persisted in WebSphere Application
Server, Version 5 only work with objects that have an unchanged remote interface.
If the remote interface is changed, the handle is no longer valid because
the stub is serialized inside the handle and its serial Version UID changes
if the remote interface changes.
This release introduces a new handle
persistence mechanism that avoids the implementation drawbacks of the previous
version. However, if handles are used for this WebSphere Application Server
deployment, you should consider the following issues when applying this update,
future WebSphere Application Server Fix Packs and EJB Container cumulative
fixes for WebSphere Application Server, Version 5.
If a WebSphere Application
Server, Version 5 persisted handle or home handle is encountered by a WebSphere
Application Server, Version 5.0.1 system, it can be read and utilized. In
addition, it will be converted to WebSphere Application Server, Version 5.0.1
format if it is re-persisted. The WebSphere Application Server, Version 5.0.1
format cannot be read by a WebSphere Application Server, Version 5 system
unless PQ72184 is applied.
Problems arise when handles are persisted
and shared across systems that are not at the WebSphere Application Server,
Version 5.0.1 level or later. However, a Version 5 system can receive a handle
from Version 5.0.1 remotely through a call to get a handle on an enterprise
bean or a getHomeHandle on an EJBHome. The remote call will succeed,
however, any attempt to persist it on the Version 5 system will have the same
limitations regarding the use of ObjectInputStream and changes in remote interface
invalidating the persisted handle.
When your application stores handles
persistently and shares this persistence with multiple clients or application
servers, apply WebSphere Application Server, Version 5.0.1 or PQ72184 to both
the client and server systems at the same time. Failure to do so can result
in the inability of these systems to read the handle data stored by upgraded
systems. Also, handles stored by the WebSphere Application Server, Version
5 can force the applications of the updated system to still subclass ObjectInputStream.
Applications using the WebSphere Application Server Enterprise, Version 5
scheduler and process choreographer, are affected by these changes. These
users should update their Version 5 systems at the same time with either Version
5.0.1 or PQ72184.
If the applications store handles in the session context,
or locally in a file on the same system, that is not shared by other applications,
on different systems, they might be able to update their systems individually,
rather than all at once. If Client Container and thin client applications
do not share persisted handle data, they can be updated as needed as well.
However, handles created and persisted in WebSphere Application Server, Version
5, Version 4.0.3 and later (with the property flag set), or Version 3.5.7 and
later (with the property flag set) are not usable if either the home or the
remote interface changes.
If any WebSphere Application Server, Version
3.5.7 or Version 4.0.3 and later enables the system property com.ibm.websphere.container.portable
to true, any handles to objects on that server have the same interoperability
limitations. In addition, if any WebSphere Application Server, Version 3.5.7
and later or Version 4.0.3 applications store a handle obtained from a WebSphere
Application Server, Version 5 or Version 5.0.1, the same restrictions apply,
regarding the need to subclass ObjectInputStream and the usability of handles
after a change to the remote interface is made.
Secure interoperability between WebSphere Application
Server, Version 5 and WebSphere Application Server, Version 5.0.1 fails when
the Lightweight Directory Access Protocol base distinguished name in not correctly
normalized in the WebSphere Application Server, Version 5 configuration
In
WebSphere Application Server, Version 5.0.1, the distinguished name is normalized
according to the Lightweight Directory Access Protocol (LDAP) specification.
In WebSphere Application Server, Version 5, the normalization of the distinguished
name is not done. The normalization consists of removing spaces in the base
distinguished name before or after commas and equal symbols.
An example
of a non-normalized base distinguished name is "o = ibm, c = us" or "o=ibm,
c=us". .
An example of a normalized base distinguished name is "o=ibm,c=us".
It
is preferred that the distinguished name is manually normalized when you enter
the base distinguished name in the configuration. In WebSphere Application
Server, Version 5.0.1 and higher version, the normalization occurs automatically
at the run time.
When a WebSphere Application Server, Version 5.0.1
system, which always has a normalized distinguished name, sends a security
token to a WebSphere Application Server, Version 5 system that contains a
non-normalized distinguished name, the request is rejected due to the mismatched
distinguished names during authorization. A common example of a request that
fails is when a WebSphere Application Server, Version 5 node is synchronizing
the configuration from a WebSphere Application Server, Version 5.0.1 deployment manager.
Another example is that a request fails when a WebSphere Application Server,
Version 5.0.1 servlet or enterprise bean makes a downstream request to a WebSphere
Application Server, Version 5 enterprise bean. However, a request going from
a WebSphere Application Server, Version 5 system to a WebSphere Application
Server, Version 5.0.1 system does not fail.
To ensure the interoperability
between WebSphere Application Server, Version 5 and Version 5.0.1, manually
normalize the base distinguished name in the LDAP configuration on a WebSphere
Application Server, Version 5.0 system by removing all the spaces before and
after the commas and equal symbols.
Internationalization service interoperability
issues between WebSphere Application Server, Version 5 and Version 5.0.1
The
WebSphere Application Server Enterprise, Version 5 internationalization service
has an interoperability problem that affects how dependent services and customer
applications behave when deserializing internationalization context across
interoperating WebSphere Application Server, Version 5 and Version 5.0.1 products.
Internationalization
context created by the internationalization service causes an InvalidClassException
when deserialized within the WebSphere Application Server runtime environment.
This is true whenever:
- Internationalization context created by Version 5 and deserialized within
the V5.0.1 runtime environment.
- Internationalization context is created by Version 5.0.1 and deserialized
within the V5 runtime environment.
Because internationalization service documentation does not instruct
customers towards the manual serialization or deserialization of internationalization
context, you should rarely, if ever, directly encounter this problem within
your applications. However, it can manifest indirectly when interoperating
between Version 5 and Version 5.0.1 releases of WebSphere Application Server
Enterprise services that process Java 2 Platform, Enterprise Edition (J2EE)
contextual information. The following services are affected by interim fix
PQ73371:
- Internationalization
- Asynchronous beans
- Scheduler
- Process choreography
To categorically avoid this problem, install the interim fix PQ73371
in all Version 5.0.1 server environments. The following section describes
the specific problematic behaviors.
Problem description: internationalization
When
attempting to deserialize internationalization context imported on an incoming
EJB (RMI/IIOP) method request between interoperating Version 5 and Version
5.0.1 releases of WebSphere Application Server Enterprise, the internationalization
service displays a message indicating an unexpected exception, InvalidClassException,
and then scopes the default locale and time zone of the local Java virtual
machine (JVM) as the caller internationalization context of the EJB method.
Because the internationalization application programming interface (API) provides methods
to set invocation internationalization context, not caller internationalization
context, this behavior is immutable.
This problem appears whenever the
following three conditions exist:
- An EJB method request is invoked between interoperating Version 5 and
Version 5.0.1 releases of WebSphere Application Server Enterprise.
- The internationalization service is enabled in the invoking client or
server environment.
- The internationalization service is enabled in the target server environment.
To avoid this problem, install the interim fix for PQ73371 to all
Version 5.0.1 server environments.
Problem description: asynchronous
beans
Asynchronous beans allow taking snapshots of Java 2 Platform,
Enterprise Edition (J2EE) contextual information from the current servlet
or EJB method. Such a snapshot is taken when using the WorkManager.create(Work
r) method to obtain an instance of WorkWithExecutionContext - a serializable
object that can be stored for future use. An instance of WorkWithExecutionContext
contains internationalization context whenever:
- The WorkManager resource used to create the instance has the com.ibm.ws.i18n context
policy attribute enabled, and
- The internationalization service is enabled.
Due to the internationalization service interoperability problem described
above, deserializing WorkWithExecutionContext objects containing internationalization
context created on the other version fails with an SerialDeserialException.
This is true for:
- Version 5 WorkWithExecutionContext objects containing Internationalization
context are deserialized on V5.0.1.
- Version 5.0.1 WorkWithExecutionContext objects containing Internationalization
context are deserialized on V5.
In the event that asynchronous beans serialization mechanisms process
an inoperable internationalization context within a WorkWithExecutionContext,
and that interim fix PQ72742 is installed in the server environment, asynchronous
beans display a message in the service logs that indicates an unexpected exception
occurred involving the internationalization service:
ASYN9999E: Unexpected Exception Occurred: com.ibm.websphere.asynchbeans.SerialDeserialException: Exception while deserializing a saved service. Service=com.ibm.ws.i18n.
When an asynchronous beans WorkManager begins executing a WorkWithExecutionContext
instance with an invalid or missing internationalization context, and the
internationalization service is enabled, the asynchronous beans scopes the
default locale and time zone of the host JVM to the Work rather than the current
invocation internationalization context.
To avoid this problem, install
the interim fix for PQ73371 to all Version 5.0.1 server environments, and
follow the directions prescribed in section PQ72742.RN.
Problem description:
scheduler
The scheduler service employs the asynchronous beans serialization
mechanisms when persisting a newly created task, and hence, is likewise affected
by the interoperability problem involving internationalization context within
the asynchronous beans WorkWithExecutionContext object.
While preparing
to run a task, if the scheduler service (that is, asynchronous beans serialization)
fails to deserialize the internationalization context, and the internationalization
service is enabled, the scheduler service scopes the default locale and time
zone of the host JVM to that task rather than the current invocation internationalization
context. See "Problem description: asynchronous beans" for error messages
and further details.
To avoid this problem, install the interim fix
for PQ73371 to all Version 5.0.1 server environments, and follow the directions
prescribed in section PQ72742.RN.
Problem description: process choreography
The
process choreographer service employs the asynchronous beans serialization
mechanisms when creating and executing interruptible flows. If a flow is
unable to deserialize the internationalization context, and the internationalization
service is enabled during execution, both compensating tasks and the flows
themselves incorrectly use the default locale and time zone of the local JVM
rather than the current invocation internationalization context. See "Problem
description: asynchronous beans" for error messages and further details.
To
avoid this problem, install the interim fix for PQ73371 to all Version 5.0.1
server environments, and follow the directions prescribed in section PQ72742.RN.
J2EE resources
Misleading scope when installing RAR file at the
server level
When installing a RAR file, by clicking Resources > Resource
Adapters > Install RAR, the scope you define on the Resource
Adapters page has no effect on where the RAR file gets installed. You can
only install RAR files at the node level.The scope on the Install RAR page
determines the node to install. However, the scope you set on the resource
adapters panel determines the scope of the new Resource Adapters, which you
can install at the server, node, or cell level.
Embedded Java Messaging Service provider installation
fails
It is possible for the embedded Java Messaging Service (JMS)
provider installation to fail without any visible warning. This situation
can occur as a result of the JMS provider prerequisite checker returning an
error that the IBM WebSphere Application Server installation is not expecting.
The IBM WebSphere Application Server installation completes, but the JMS provider
is not installed.
If the JMS installation fails, look for details in
the mq_prereq.log file in the /tmp directory and in the create_mq.log and mq_install.log files
in the $WAS_HOME$/log directory. When you have corrected the situation,
install the embedded JMS provider separately by clicking the Custom installation then
selecting either one or both of the following options, as required: Embedded
Messaging Server and Embedded Messaging Client. For more information
about installing the embedded JMS provider, see the InfoCenter article "Installing
WebSphere embedded messaging as the JMS provider".
In the single server environment, the embedded
Java Messaging Service server and IBM WebSphere Application Server run on
the same single Java virtual machine
The embedded Java Messaging
Service (JMS) server and the IBM WebSphere Application Server run on the same
single Java virtual machine. Due to this architecture, if the MQ process (such
as amq*.exe - not JMS process but MQ native proces) dies and restarts while
the JMS server is running, you must restart the IBM WebSphere Application
Server (ordinarily named "server1") manually. This shutdown also can affect
other installed applications.
In the distributed server environment,
the JMS server and the application server run in separate processes. Therefore,
the JMS server is not required to restart the application server.
To
work around this problem, restart the application server (server1).
Selecting the Java Messaging Service resources
when deploying the WAR file
When you deploy a WAR file and select
the JMS resources that you defined, you receive a warning message without
selecting the JMS resources.
Information you should know about using
server-side and client-side selectors
The default behavior for the
internal JMS broker is to use server-side selectors. The default behavior
for external brokers is to use client-side selectors. The reason for the latter
is that not all brokers support server-side filtering. Those that do, are
not all JMS compliant in their implementation process.
Stopping the queue manager while running
the embedded Java Messaging Service provider stops the IBM WebSphere Application
Server
When you run the embedded Java Message Service (JMS) provider
and the queue manager stops, IBM WebSphere Application Server is also stopped.
To
recover the queue manager and IBM WebSphere Application Server, you must start
the application server.
Limited connection caching available
for servlets
Caching of connection handles across servlet
methods is limited to JDBC and Java Message Service (JMS) resources. Other
non-relational resources, such as Customer Information Control System (CICS)
or IMS, currently cannot have their connection handles cached in a servlet.
This limitation only applies to single-threaded servlets since multithreaded
servlets do not allow caching of connection handles.
To work around
this problem, you need to get, use, and close the connection handle within
each method invocation.
Problems with Xerces versions that are not J2EE
compliant
Programs containing the Xerces parser routine that work
on IBM WebSphere Application Server, Version 4.0.x might not work on IBM WebSphere
Application Server, Version 5.
Versions of Xerces prior to 2.0.0, including
the versions shipped with IBM WebSphere Application Server, Version 4.0.x,
were not J2EE1.3 compliant. To meet J2EE1.3 certification constraints (DOM2/SAX2/JAXP1.1
(and no higher)), IBM WebSphere Application Server, Version 5 requires XML4J4.0.0
(Xerces 2.0 bug fixes).
In Xerces versions prior to 2.0.0 and Xalan
versions prior to 2.2, the DOM, SAX, and JAXP APIs were bundled inside of
the xerces.jar and xalan.jar files.
In the most recent versions, these non-Apache XML APIs have moved into a separate
JAR file in the XML common apache project. These common XML APIs are required
by the specifications and contained within the IBM WebSphere Application Server,
Version 5 j2ee.jar file.
The cause of the
problem is that the class loader behavior mixes the classes that are required
by J2EE1.3 with those of the older Xerces implementation.
To work around
this problem, use the version that ships with IBM WebSphere Application Server,
Version 5.
National Language Version issues and limitations
Patches needed for correct character display
of the installation panel on Asian language platforms
If the installation panel displays garbled characters
on Asian language platforms, including Korean, traditional Chinese, Japanese,
and simplified Chinese, on the HP-UX 11i platform, make sure that you have
the following patches installed on your HP-UX 11i machine:
| Name |
Brief Description |
| B6848BA |
Ximian GNOME 1.4 GTK+ Libraries for HP-UX 11.00 and
11i |
| BUNDLE11i |
Required Patch Bundle for HP-UX 11i, February 2001 or
later |
| PHSS_26974(for 11.0) or PHSS_26973 (for 11i) |
Korean TrueType fonts |
| PHSS_24937(for 11.0) or PHSS_26977 (for 11i) |
Traditional Chinese TrueType fonts |
| PHSS_26972(for 11.0) or PHSS_26971 (for 11i) |
Japanese TrueType fonts |
| PHSS_26976(for 11.0) or PHSS_24975 (for 11i) |
Simplified Chinese TrueType fonts |
| PHSS_25091(for 11.0) or PHSS_25092 (for 11i) |
Common TrueType fonts |
These patches can be found at the following URL: http://www.hp.com/products1/unix/java/infolibrary/font_info.html.
Cannot display newly created Windows domain
users during application management for the non-English locales only )
The
newly created domain users are not retrieved when you perform application
management. The newly created groups are displayed correctly and the newly
created local users are also displayed correctly. If you map users to roles
by clicking A Role > Choose Users > Find Users, only users created
on Primary Domain Controller, before adding the client machine to the
domain are displayed. If a new user is created while the client machine is
already on the domain, the getUser function does not retrieve the
new users created on Primary Domain Controller. The getUser API
works correctly for the English locales but does not work correctly for the
non-English locales.
A
404 Not Found exception thrown when cells contain nodes on different locals
On
UNIX platforms, a "com.ibm.websphere.management.filetransfer.client.TransferFailedException:
File download failed [404].Message: 404 Not Found" exception is thrown when
federating nodes with non-matching double-byte character set (DBCS) locales.
All
nodes incorporated into the same cell must be located on host machines with
an identical locale setting. Cells containing nodes on different locales do
not function properly and are not supported.
Help search does not return hits for Latin-1
terms with diacritical markings
The help search in the administrative
console does not return search results for Latin-1 terms that contain diacritical
markings (for example, German "fur", where the "u" includes an umlaut). To
work around this problem, use the decimal representation of the character
in the search term (for example, "für").
Cannot access readme and PDF installation files
for Traditional Chinese and German languages through the LaunchPad
You
can view these files through any Web browser.
The readme file for Traditional
Chinese is located at:
<cd drive>\readme\readme_zh_tw.html
The
installation PDF file for Traditional Chinese is located at:
<cd drive>\doc\installguide_zh_tw.pdf
The
readme file for German is located at:
<cd drive>\readme\readme_de.html
The
installation PDF file for German is located at:
<cd drive>\doc\installguide_de.pdf
Avoid missing characters in license agreement
Before
installing the IBM WebSphere Application Server on a Traditional Chinese AIX
machine, type the following to avoid missing characters within the license
agreement:
cd /usr/lib/nls/loc/iconv/
mv IBM-eucTW_CNS11643.1986-1 IBM-eucTW_CNS11643.1986-1.save
mv big5_CNS11643.1986-1 big5_CNS11643.1986-1.save
Help links for Application Assembly Tool, Application
Resource Client Tool and Tivoli Performance Viewer bring up the English files
To
work around this problem, change the path to the preferred language version
to display the appropriate language (include list for languages if possible)
list separately if the paths are not the same.
Product registration displays as English when
installing in Brazilian Portuguese
When registering your product
from the installation in Brazilian Portuguese, the product registration page
displays in English.
To view the registration page in Brazilian Portuguese,
you can enter the following URL, <install_root>\prt\PRT_Welcome_ptb.html or
register the product using FirstSteps in Brazilian Portuguese.
Running servers
in different code pages within a single cell is not supported
Both
the IBM WebSphere Application Server Network Deployment node and the IBM WebSphere
Application Server nodes must be on the same code page.
The Java virtual machine can not translate coded
Character Set Identifier used by the embedded messaging queue manager
When
running the IBM WebSphere Application Server with global security enabled
the embedded messaging provider can fail to start with message MSGS0504E and
an UnsupportedEncodingException:
MSGS0504E: JMS Server Security Service Thread received Socket Exception:
java.io.UnsupportedEncodingException: Cp5050
This message
indicates that the Java virtual machine (JVM) cannot translate the Coded Character
Set Identifier (CCSI) used by the embedded messaging queue manager.
To
work around this problem, set a locale and language value that maps to a code
page supported by the JVM. Refer to the WebSphere MQ messaging platform-specific
books Web site at http://www-3.ibm.com/software/ts/mqseries/library/manualsa/manuals/platspecific.html
for
a "WebSphere MQ Platform Quick Beginnings V5.3" manual that is relevant to
your platform. Look for a table in the manual with a list of locale and language
to CCSI mappings supported by the queue manager. Then, select a locale and
language that maps to a CCSID that is supported by the JVM. Set the selected
locale and language, and restart the IBM WebSphere Application Server.
Insufficient prerequisites cause warning messages
during installation with AIX 4.3.3
When
installing the Simplified Chinese version of the IBM WebSphere Application
Server on AIX 4.3.3, warning messages display because of insufficient prerequisites.
The error message indicates that the file levels X11.fnt.ucs.ttf, X11.fnt.ucs.ttf_TW, and X11.fnt.ucs.ttf_KR do
not match the level of file X11.fnt.ucs.ttf_CN.
The latter file has a level of 4.3.3.25, while the others have the level 4.3.3.
To
work around this problem, force the installation by modifying the prereqChecker.xml file
to the required file level.
Japanese not recommended on Red Hat Advanced Server
Version 2.1
The Red Hat Advanced
Server Version 2.1 GUI language selection menu displays two selections for
Japanese: Japanese (eucJP) and Japanese (SJIS).
Click Japanese
(eucJP) to install and run the IBM WebSphere Application Server in a Japanese
environment. Using Japanese (SJIS) is not recommended.
For reference
see the Red Hat Japanese Web site at http://www.jp.redhat.com/support/7.2/sla/rh72faq.html
.
Evaluating the product on National Language Version
operating systems
The following steps are work arounds to evaluate IBM WebSphere
Application Server and IBM WebSphere Application Server Network Deployment,
Version 5 on National Language Version operating systems.
- Windows NT platform:
- Open Regional Settings from the Windows NT Control Panel.
- Change the setting to English (United States) on Regional Settings,
and then select set as system default locale check box.
- Add English (United States) on Input Locales, and set the
default input locale to English.
- Click OK on the Regional Settings Properties window.
- Restart the machine with the English locale settings.
- Install the IBM WebSphere Application Server, Version 5.
Note: Do not change locale settings to original non-English settings
until you finish the evaluation.
- Windows 2000 platform:
- Open Regional Options from the Windows 2000 Control Panel.
- Change the setting Your locale (location): to English (United
States) on General, click Default Settings, and then change
the system locale to English (United States).
- Add English (United States) on Input Locales, and set the
default input locale to English.
- Click OK on the Regional Options window.
- Restart the machine with the English locale settings.
- Install the IBM WebSphere Application Server, Version 5.
Note: Do not change locale settings to original non-English settings
until you finish the evaluation.
- AIX 4.3.3 or AIX 5.1:
Open Terminal and run the following command
before you install the IBM WebSphere Application Server, Version 5:
# export LANG=en_US
Note: Every time that
you open a new terminal, you need to invoke the above command.
- Solaris Operating Environment 8:
Open Terminal and run the following
command before you install the IBM WebSphere Application Server, Version 5:
# LANG=en_US
# export LANG
Note: Every time you open a new terminal, you need
to invoke the above command.
Performance data and tools
Enabling Performance Monitoring Infrastructure in
NodeAgent
Performance Monitoring Infrastructure (PMI) is disabled
by default on NodeAgent. To enable PMI on NodeAgent:
- Open the administrative console.
- Click System Administration > Node Agents > nodeagent > Performance
Monitoring Service.
- Select the check box in the Startup field.
- Click Apply or Ok.
- Click Save.
- Restart NodeAgent.
How to run a Performance Monitoring Infrastructure
application
The article entitled, "Running your new monitoring
applications" shows an incorrect script for using WSLauncher. The correct
script follows:
call "~dp0setupCmdLine.bat"
set WAS_CP=%WAS_HOME%\lib\properties
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\pmi.jar
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\pmiclient.jar
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\ras.jar
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\admin.jar
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\wasjmx.jar
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\j2ee.jar
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\soap.jar
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\soap-sec.jar
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\nls.jar
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\wsexception.jar
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\ws-config-common.jar
%JAVA_HOME%\bin\java "%CLIENTSOAP%" "%CLIENTSAS%" "-Dws.ext.dirs=%WAS_EXT_DIRS%"
%DEBUGOPTS% -classpath "%WAS_CP%" com.ibm.websphere.pmi.PmiClientTest host name [port] [connectorType]
More information about RangeStatistic data value
The
article entitled, "Performance data classification" is missing the RangeStatistic
data current value. RangeStatistic data contains current values, as well as
lowWaterMark and highWaterMark.
More information about total memory counter in
Java Virtual Machine run-time module
The total memory counter in
the Java Virtual Machine (JVM) data category is a BoundedRangeStatistic type.
However, the upperBound and lowerBound are not implemented in this release.
How to run a Performance Monitoring Infrastructure
client application when security is enabled
To run a PMI client
application, such as Tivoli Performance Viewer, when security is enabled,
make sure that you have %CLIENTSOAP% and %CLIENTSAS% properties
on your JVM command line. The %CLIENTSOAP% and %CLIENTSAS% properties are defined in the setupCmdLine.bat or setupCmdLine.sh file.
It is not necessary to
change the PMI application code when security is enabled but you have to edit
the property file WAS/properties/soap.client.props when
the Simple Object Access Protocol (SOAP) connector is used. In the soap.client.props file,
set com.ibm.SOAP.securityEnabled to true.
Also set com.ibm.SOAP.loginUserid and com.ibm.SOAP.loginPassword to
the user ID and password for login.
You have to edit the soap.client.props file
as described here for the SOAP connector. For RMI connector, you can either
edit the sas.client.props file or type in the
user ID and password through a pop-up window if you do not put them in the
property file. Note: A common mistake is leaving extra spaces at the end of
the lines in the property file. Do not leave extra spaces at the end of the
lines especially for the user ID and password lines.
Information you should know about JDBC data source
names in Performance Monitoring Infrastructure
Performance Monitoring
Infrastructure (PMI) collects performance data for both 4.0 and 5.0 JDBC data
sources, which you can find under JDBC Connection Pool data contents.
For a 4.0 data source, the data source name is used. For a 5.0 data source,
the Java Naming and Directory Interface (JNDI) name is used.
Better startup performance requires a minimum heap
size of 50MB
For better startup performance, it is recommended that
you set minimum heap size to 50MB, that is, -Xms50m.
Samples
All platforms
WebSphere Application Server,
Version 5.0.1 Samples installation and instructions
WebSphere Application
Server, Version 5.0.1 contains updates for the Samples. The updates contain
a number of minor fixes to the original Samples that are shipped with the
product. The Samples documentation update is located in the samples50_fp1 directory.
This directory is located off of the root directory of where WebSphere Application
Server is installed. The file name is readme.pdf. The document
contains instructions for uninstalling the Version 5 Samples and reinstalling
the Version 5.0.1 Samples.
Samples Gallery font in languages other than English
can appear small
The Samples Gallery font in languages other than
English can appear small, depending on the browser you use. To work around
this problem, increase the browser font settings.
Coexistence when using
Point-to-Point and Publish and Subscribe Samples
During installation,
it is possible to change the default ports for the bootstrap and Simple Object
Access Protocol (SOAP) connector ports. If these ports are changed, additional
parameters are required to override the default ports when using the launchclient command
for the Point-to-Point and Publish and Subscribe Samples. The current documentation
for the Samples assumes the default port settings. The additional parameters
are:
-CCBootstrapPort=<bootstrap port>
-CCsoapConnectorPort=<soap connector port>
Creating new
account causes duplicate key exceptions
You can create new accounts
several ways in Pet Store. If Account is selected from a Pet Store
screen before you signed in, you have the option of signing in, or creating
a new account. If a new account is created in this flow, it fails with duplicate
key exceptions. To create a new account click Sign in from a Pet Store
screen.
Passwords
are not compared when creating a new account
When creating a new
account, you must provide a new password. The first Password field
is not compared for a match with the second Password (Repeat) field.
Account information
displays incorrectly
Your account information displays when Account is
selected, or during the checkout process. First and last names are switched,
the credit card name shows the expiration date, and expiration date shows
the wrong date.
Items are
added to the shopping cart with the wrong price
The Pet Store Item
screen shows a picture of the product to purchase with a list price and your
price. The list price is greater than your price. Items added to the shopping
cart are added with the more expensive list price.
Add to cart resets
the quantity of an item to one
When an item is added to the shopping
cart it is added with an initial quantity of one. You can adjust the quantity
to a higher number in the shopping cart. If that same item is added again
to the shopping cart, the adjusted quantity is reset to one.
Pet Store
pages do not show animal pictures or details when HTTP Transport is reconfigured
to use a port other than 9080
If the Pet Store pages do not show
animal pictures or details when you reconfigure the IBM WebSphere Application
Server to use an HTTP Transport other than port 9080, the URL of Pet Store
Catalog DAOSQLURL needs updating to use the new port.
You can reconfigure
the URL resource to use the new port with the IBM WebSphere Application Server
administrative console. Go to Resources > URL Providers (set scope to
your Server) > Samples URL Provider - https > URLs > Pet Store Catalog
DAO SQL URL. The port number is in the Specification field.
Forceful repopulating
of Pet Store database fails
The option to forcefully repopulate
the Pet Store database fails with an internal server error (HTTP 500 - The
page cannot be displayed). Refrain from using this option.
$AdminApp installation documentation describes
the -deployejb.dbtype option incorrectly
The documentation for $AdminApp
installation describes the -deployejb.dbtype option as CLOUDSCAPE_V50. The
correct option is CLOUDSCAPE_V5.
The following command is an example
of its usage. The command deploys an application using IBM Cloudscape with
the correct database type:
wsadmin -conntype none -c "$AdminApp install TechnologySamples.ear
{-appname TechnologySamples -usedefaultbindings -node <node> -deployejb
-deployejb.dbtype CLOUDSCAPE_V5}"
Security
All platforms
An error message occurs in the native_stdout.log
file when enabling security on HP- UX 11i platforms
After you enable security on HP-UX 11i platforms, the
following error in the native_stdout.log file occurs, along with
a core dump and WebSphere Application Server does not start:
Java HotSpot(TM) Server VM warning: Unexpected Signal 11 occurred under user-defined signal handler 0x7895710a
To work around this error, apply fixes recommended by HP for Java
at the URL: http://www.hp.com/products1/unix/java/infolibrary/patches.html.
getCallerPrincipal() returns a realm/realm/user format
when you specify a Domain Qualified User Name
When you specify Domain
Qualified User Name from the Security > Global Security configuration
panel, the runtime call to getCallerPrincipal() from an enterprise
bean returns the qualified name with the realm prepended twice. For example,
the format return is realm/realm/user. You can strip the first realm
from the returned value when making API calls. The servlet API getUserPrincipal() works
correctly.
Test connection fails for data sources with
newly added J2C authentication data alias
If you create or update
a data source that points to a newly created J2C authentication data alias, Test
Connection fails to connect until you have restarted the deployment manager
so that the J2C authentication data is reflected in the run-time configuration.
Any changes to the J2C authentication data fields require a deployment manager
restart for the changes to take effect.
WebSphere Application Server security help
files incorrectly mapped
Some of the links from the following WebSphere
Application Server security windows to their corresponding help files are
incorrectly mapped:
Environment > Naming > CORBA Naming Service
Users
Environment > Naming > CORBA Naming Service Users > Add
Environment
> Naming > CORBA Naming Service Groups
Environment > Naming >
CORBA Naming Service Groups > Add
System Administration > Console
Groups > Add
Note: In the System Administration > Console
Groups > Add window, the links to the help located next to Group Description and Role
Description are correct. However, the link to the help text at the top
of the help window is incorrect.
If you access a help file that does
not correspond to the appropriate WebSphere Application Server window, use
the following table to locate the correct help information. When you access
the incorrect help file, find the listing of help files located under Core
Console on the left side of your current help file window and click the
appropriate link.
| Window path |
Correct Help file |
| Environment > Naming > CORBA Naming Service Users |
CORBA Naming Service users settings |
| Environment > Naming > CORBA Naming Service Users
> Add |
CORBA Naming Service users settings |
| Environment > Naming > CORBA Naming Service Groups |
CORBA Naming Service groups |
| Environment > Naming > CORBA Naming Service Groups
> Add |
CORBA Naming Service groups |
| System Administration > Console Groups > Add |
Console groups settings |
Java Authentication and Authorization
Service configuration setting for Java client program
You must invoke
Java client programs that use Java Authentication and Authorization Service
(JAAS) for authentication with a JAAS configuration file specified. The WebSphere
product supplies the default JAAS configuration file, wsjaas_client.conf under
the <install_root>/properties directory.
This configuration file is set in /<install_root>/bin/launchClient.bat file
as:
set JAAS_LOGIN_CONFIG=-Djava.security.auth.login.config=%WAS_HOME%\properties\wsjaas_client.conf
If launchClient.bat file is not used to invoke Java client
programs, make sure the appropriate JAAS configuration file is passed to the
Java virtual machine with the
-Djava.security.auth.login.config flag.
Default Secure Sockets
Layer alias of the cell is used to create HTTPS transport
On the
IBM WebSphere Application Server Network Deployment installation, the Secure
Sockets Layer alias of the default cell is used for the HTTPS transport when
creating a new server.
The javax.security.auth.Subject.getSubject(java.security.AccessControlContext)
method does not work within the AccessController.doPrivileged() method
You
can retrieve the subjects in a Subject.doAs() block with the Subject.getSubject()
call. However, this procedure does not work if there is an AccessController.doPrivileged()
call within the Subject.doAs() block. In the following example, s1 is
equal to s, but s2 is
null:
AccessController.doPrivileged() not only truncate the Subject propagation, also reduce the
permissions, it does not includes the JAAS security policy defined for the principals in the Subject.
Subject.doAs(s, new PrivilegedAction() {
public Object run() {
System.out.println("Within Subject.doAsPrivileged()");
Subject s1 = Subject.getSubject(AccessController.getContext());
AccessController.doPrivileged(new PrivilegedAction() {
public Object run() {
Subject s2 = Subject.getSubject(AccessController.getContext());
return null;
}
});
return null;
}
});See the documentation for information on how to use the
WSSubject class instead of the Subject object to associate IBM WebSphere Application
Server credentials to the thread of context.
The IBM Java Secure Socket Extension is not
supported within applets
The IBM Java Secure Socket Extension (JSSE)
is currently not supported within applets.
The shadow password file must exist for the local
operating system security to work
For the IBM WebSphere Application Server local operating system
security registry to work on Linux systems and a Solaris Operating Environment,
a shadow password file must exist. The shadow password file is named shadow
and is in the /etc directory. If the shadow password
file does not exist, an error occurs after enabling global security and configuring
the user registry as local operating system.
To create the shadow file,
run the pwconv" command (with no parameters). This command creates
an /etc/shadow file from the /etc/passwd file.
After creating the shadow file, you can configure local operating system security.
The grant entries in the app.policy and was.policy files must have a code base
Grant
entry specified in the app.policy and was.policy files
must have a code base defined. If there are grant entries specified without
a code base, the policy files are not loaded properly and the application
can fail.
If the intent is to grant the permissions to all applications,
then use file:${application} as a code base in
the grant entry.
SecurityManager does not check modifyThread
and modifyThreadGroup permission for non-system threads
When Java
2 Security is enabled in the Global Security settings, the installed SecurityManager
does not currently check modifyThread and modifyThreadGroup permissions for
non-system threads.
You can use Web applications and EJB application
code to create threads and access to the following thread class methods: stop,
suspend, resume, setPriority, setName, and setDaemon.
Allowing Web and
EJB application code to create or modify a thread can have a negative impact
on other components of the container and can affect the capability of the
container to manage enterprise bean life cycles and transactions.
Information you should know about the Java Authentication
and Authorization Service login configuration updates in the Security Center
The
Java Authentication and Authorization Service (JAAS) login configuration entries
in the Security Center are propagated to the server run time when they are
created, not when the configuration is saved.
However, the deleted JAAS
login configuration entries are not removed from the server run time. To remove
the entries, save the new configuration, then stop and restart the server.
Web client certificate authentication is not currently
supported with the local operating system authentication mechanism
Web
client certificate authentication is not currently supported when using the
local operating system (LocalOS) user registry. However, Java client certificate
authentication does work with LocalOS. Java client certificate authentication
maps the first attribute of the certificate domain name to the user ID in
the LocalOS user registry.
Even though Java client certificates work
properly, the following error displays in the SystemOut.log file:
SECJ0337E: The mapCertificate method is not supported
The error is intended for Web client certificates, but also displays
for Java client certificates. You can ignore this error for Java client certificates.
Identity assertion fails when basic authentication
is not selected
When specifying identity assertion on the CSIv2
Authentication Outbound panel, you must also select basic authentication as
supported or required on the CSIv2 Authentication Outbound panel. This action
allows the server identity to be submitted, along with the identity token,
so that the receiving server can trust the sending server. Without
specifying basic authentication as supported or required, trust is not established
and the identity assertion fails.
Access denied
in mixed security configuration
If your security configurations
are not consistent across all servers, you get access denied errors.
When
enabling or disabling global security in a Network Deployment environment,
it is important that your new security setting is propagated to all the nodes
before restarting the deployment manager and node agents to change the new
security policy.
Configuration changes such as this are generally propagated
using configuration synchronization. If auto-synchronization is enabled, you
can wait for the automatic synchronization interval to pass or you can force
synchronization before the synchronization interval expires. If you use manual
synchronization, you must synchronize all nodes.
If the cell is in a
configuration state, where the security policy is mixed with nodes that have
security enabled and nodes that have security disabled, you can use the syncNode
utility to synchronize the nodes where the new settings were not propagated.
Refer
to the article entitled, "Enabling and removing global security" in the Network
Deployment version for more detailed information about enabling security in
a distributed environment.
Do not use forward slash character in the JAAS
login configuration alias name
Do not use the forward slash character
(/) in the alias name when defining JAAS login configuration entries. The
JAAS login configuration parser cannot handle the forward slash character.
Exposing the JAAS Subject object and WSPrincipal
interface
Several release notes follow that relate to how IBM WebSphere
Application Server, Version 5 exposes the JAAS Subject object and WSPrincipal
interface.
- The Subject object that is generated by the WSLoginModuleImpl instance
and WSClientLoginModuleImpl instance contains a principal that implements
the WSPrincipal interface. The getCredential() method of a WSPrincipal object
returns an object that implements the WSCredential interface. You can also
find the WSCredential object instance in the PublicCredentials list of the
subject instance. It is recommended that you retrieve the WSCredential object
from the PublicCredentials list instead of using the getCredential() method.
- The getCallerPrincipal() method of the WSSubject class returns a String
representing the caller security identity. Note that the return type is different
from that of the getCallerPrincipal method of the EJBContext interface, which
is java.security.Principal. The article entitled, "Java 2 Connector security",
under the heading "J2C mapping module programming reference", incorrectly
states that the WSSubject getCallerPrincipal() method returns a com.ibm.websphere.security.auth.WSPrincipal
object. Incorrect information is also in the related article entitled, "Developing your
own J2C principal mapping module."
- The Subject object generated by the J2C DefaultPrincipalMapping module
contains a resource principal and a PasswordCredentials list. In the present
implementation the resource principal represents the caller. See the article
entitled, "Java 2 Connector security."
NO_PERMISSION returned during lookup request
If
a NO_PERMISSION error is returned to the IBM WebSphere
Application Server, Version 5 client during a lookup request, apply PQ67441
for IBM WebSphere Application Server for z/OS, Version 4.0 to securely interoperate
with an IBM WebSphere Application Server, Version 5 client.
The location of the SOAP security properties file
is incorrect in the setupClient.bat and setupClient.sh files
Note: This section applies to the application clients installation
of WebSphere Application Server, Version 5 Fix Pack 1.
The setupClient.bat file
on Windows platforms and the setupClient.sh file on UNIX platforms
incorrectly specify the location of the SOAP security properties file.
In
the setupClient.bat file, the correct location should be:
set CLIENTSOAP=-Dcom.ibm.SOAP.ConfigURL=file:%WAS_HOME%/properties/soap.client.props
In the setupClient.sh file, the CLIENTSOAP variable
should be:
CLIENTSOAP=-Dcom.ibm.SOAP.ConfigURL=file:$WAS_HOME/properties/soap.client.props
In
the setupClient.bat and setupClient.sh files, complete the
following steps:
- Remove the leading / after file:.
- Change sas to soap.
Web services
Preserving an existing
Web Services Gateway configuration when installing the fix pack
The
update installer installs the enterprise applications for the Web Services
Gateway in the /installableApps directory of WebSphere Application
Server Network Deployment. To deploy the enterprise applications across the
cell, complete the following steps:
- Back up the following gateway configuration files to a secure location,
for later restoration. You can find these files on each application server
on which the gateway has been installed.
- WAS_HOME /installedApps/node_name/wsgwsoap1.ear/wsgwsoap.war/DeployedServices.ds
- WAS_HOME/installedApps/node_name/wsgwsoap1.ear/wsgwsoap.war/wsgwsoap.channelname.cfg
- WAS_HOME/installedApps/node_name/wsgwsoap2.ear/wsgwsoap.war/DeployedServices.ds (if
installed)
- WAS_HOME/installedApps/node_name/wsgwsoap2.ear/wsgwsoap.war/wsgwsoap.channelname.cfg (if
installed)
- WAS_HOME/installedApps/node_name/wsgw.ear/wsgw.war/wsgw.cfg
where WAS_HOME is the root directory for each installation
of the application server (by default WebSphere/AppServer).
- Install all fix packs for all WebSphere Application Server products by
following the instructions in the product documentation article: "Installing
fixes and fix packs."
- Use the WebSphere Application Server administrative console to stop and
remove the following enterprise applications:
- wsgw
- wsgwsoap1
- wsgwsoap2 (if installed)
- Complete the installation of the gateway by following the instructions
in the product documentation article: "Web Services Gateway - Completing the
installation."
- Stop the wsgw, wsgwsoap1, and wsgwsoap2 (if installed) applications.
- Restore the gateway configuration files that you backed up in step 1,
by copying them back to their original locations. Replace any files of the
same name that were created when the enterprise applications started.
- Restart the wsgw, wsgwsoap1, and wsgwsoap2 (if installed) applications.
Installing the Web
Services Gateway into a deployment manager cell
If you are installing
to your first remote server, complete the following steps:
- Follow the instructions given in the article entitled, "Installing the
gateway into a deployment manager cell".
- Use FTP to copy the WSGWResourceBundles.jar file from the WebSphere_DeployMgr_root/lib directory
(by default WebSphere/DeploymentManager/lib) to the /lib directory
of the remote server (by default WebSphere/AppServer/lib).
For subsequent servers, complete the following steps:
- Use the IBM WebSphere Application Server Network Deployment administrative
console to map the new server to each existing gateway application (in the
administrative console click the required application, then click Map modules
to application servers, highlight the required servers and apply them
to the required modules).
- Use FTP to copy the WSGWResourceBundles.jar file
from the WebSphere_DeployMgr_root/lib directory
(by default WebSphere/DeploymentManager/lib) to
the /lib directory of the remote server.
- Stop, then restart the application servers.
Web Services Technology
Preview does not work in WebSphere Application Server, Version 5.0.1
When
the Web Services Technology Preview is installed with the Web Services Security
options selected, the Web Services invocation fails with the following errors:
Server
side exception when invoking the Web Services:
[3/13/03 11:30:28:641 CST] 7d090dc5 WebGroup E SRVE0026E: [Servlet Error]-[com.ibm.ws.security.core.SecurityContext: method getCurrent()Lcom/ibm/ISecurityLocalObjectBaseL13Impl/CurrentImpl; not found]: java.lang.NoSuchMethodError: com.ibm.ws.security.core.SecurityContext: method getCurrent()Lcom/ibm/ISecurityLocalObjectBaseL13Impl/CurrentImpl; not found
On the client side, the following default is returned:
AxisFault
faultCode: {http://xml.apache.org/axis/}HTTP
faultString: (500)Internal Server Error
faultActor: null
faultDetail:
null: return code: 500
Error 500: com.ibm.ws.security.core.SecurityContext: method getCurrentt()Lcom/ibm/ISecurityLocalObjectBaseL13Impl/CurrentImpl; not foundThe reason for these errors is that the WebSphere Application Server security
run time in WebSphere Application Server Version 5.0.1 is updated but the
Web Services Technology Preview is not.
Complete either of the following
actions to avoid receiving the errors:
- During installation of Web Services Technology Preview, do not select Web
Services Security options.
- If Web Services Technology Preview is installed with Web Services Security options,
remove the following JAR files:
${WAS_HOME}/lib/was-wssecurity.jar
${WAS_HOME}/lib/webservices-security.jarYou must stop the application
server before removing the JAR files. After the JAR files are removed, you
can invoke Web Services without selecting the Web Services Security option.
Two class libraries are
available for Universal Description, Discovery, and Integration 4J
Universal
Description, Discovery, and Integration 4J (UDDI4J) is a class library that
provides an API used to interact with a UDDI registry.
Two class libraries
are provided for UDDI4J in the library subdirectory:
- uddi4j.jar supports Version 1 of the UDDI
API. This support is provided for compatibility with applications written
using UDDI Version 1, and the classes in this class library are deprecated.
- uddi4jv2.jar supports Version 2 of the UDDI
specification. Any application using UDDI4J to communicate with a Version
2 UDDI-compliant registry should use the class in this library.
Errors occur if you stop
and restart the UDDI Registry application from the administrative console
and then use the user console
If you stop and restart the UDDI Registry
application from the administrative console, and then try to access the registry
through the user console, an Error 500 - object is
not an instance of a declaring class displays on the user console, and
the error message SRVE0026E displays in the system
log. You cannot access the UDDI Registry until you restart the IBM WebSphere
Application Server.
To avoid seeing this problem when you restart the
UDDI registry application, perform the following steps using the IBM WebSphere
Application Server Administrative Console:
- Click Applications > Enterprise Applications.
- In the list of applications, locate and click the UDDI application link.
- Scroll down to the Related Items section and click Web Modules.
- Click gui.war.
- Change the Classload Mode value to PARENT_LAST (instead
of PARENT_FIRST).
An UDIN2041I error
might occur with the wsadmin appserversetupuddi.jacl command
The
following scripting error might display when you run the wsadmin appserversetupuddi.jacl command:
UDIN2041I: Starting UDDI application. UDIN8019E: startApplication command for appname caught exception Exc. Values are: appname=UDDIRegistry, Exc=com.ibm.ws.scripting.ScriptingException: com.ibm.websphere.management.exception.ConnectorException: ADMC0009E: Failed to make the SOAP RPC call: invoke
During installation, if you see the previous error after running the appserversetupuddi.jacl command,
you can ignore the error. It is recommended that you start (or stop and restart)
the application server and then continue.
Working with HTTP servers
Cannot start IBM HTTP Server 2.0.42.1 on AIX 5.2
systems
There is a potential IBM HTTP Server 2.0.42.1 startup failure
on AIX 5.2 systems when a non-default AIX resolver configuration is used,
for example, in /etc/netsvc.conf, and the IBM HTTP Server configuration
file, for example, httpd.conf, uses the Listen directives
that do not specify an Internet Protocol address. The following is an example
of the error messages issued by IBM HTTP Server:
[crit] (78) A remote host did not respond within the timeout period.: alloc_listener: failed to set up sockaddr for ::
Syntax error on line 130 of /usr/IBMIHS/conf/httpd.conf
Listen setup failed.
This error is caused by a defect in the AIX resolver library. The
defect is fixed by APAR IY40908 for AIX 5.2. The fix is shipped with AIX52B
Gold and later.
In the meantime, you can take one of the following actions
to solve the problem:
- Use the IPv4-specific Listen directive in the IBM HTTP Server
configuration file, for example, "Listen 0.0.0.0:80"
- Do not override the default AIX resolver configuration.
- Change the following line in the /etc/netsvc.conf file from:
hosts = local = auth , bind
to
hosts = local = auth , bind , local
l
Change the permissions of four files to reduce
the risk of security exposure
When
you update the IBM HTTP Server with WebSphere Application Server, Version
5.0.1on UNIX platforms, there are four files with 777 permissions that need
to be updated. To change the permissions, issue the command chmod 644 <filename>,
where the filename is one of the following files:
<IHS install location>/example_module/Makefile.exmpl
<IHS install location>/example_module/mod_example.exp
<IHS install location>/ssl/ikeyman/en_US/sgkey.sgs
<IHS install location>/version.signature
You must change the
permissions of these files to reduce the risk of security exposures. Also,
if you follow these instructions and change the permissions, the functionality
of IBM HTTP Server is not adversely affected.
ImagePath entry not updated in registry when you
change the document root
When you change the document root in the
IBM HTTP Server Version 1.3.26 administration server, the ImagePath entry
in the registry, corresponding to the HTTP server, does not update. There
are no reported functionality problems caused by this registry entry.
Forms Proxy Settings and Proxy Cache do not behave
correctly
When using Internet Explorer Version 6.0 to view the IBM
HTTP Administration server, the Forms Proxy Settings and Proxy Cache do not
behave correctly. If you select the radio buttons that display editable fields,
the fields do not display.
To work around this problem, use another
version of the browser.
Unable to use the default service names
created by the installer because they are common
In a same version coexistence scenario for IBM HTTP Server Version
1.3.26 or Version 2.0.42 on a Windows platform, you are unable to use the
default service names created by the installer because they are common.
To
work around this problem:
- Install the first copy of IBM HTTP Server, either by itself or with IBM
WebSphere Application Server.
- Customize the service names for the first install by running the following
commands from the first install location:
apache -k install -n "IHS 1.3.26(1)"
apache -k install -f conf\admin.conf -n "IHS 1.3.26 Administration (1)"
- Edit the AdminAlias directive in the <installLocation 1>\conf\admin.conf
file to point to the new service name, for example "IHS 1.3.26(1)".
- Remove the default service names installed by the first install by running:
apache -k uninstall -n "IBM HTTP Server 1.3.26"
apache -k uninstall -n "IBM HTTP Administration 1.3.26"
- Install the second copy of IBM HTTP Server, either by itself or with IBM
WebSphere Application Server. The default service names will now correspond
to the second install.
Note: The custom service names above need to be unique on your system
and are not required to be those exact strings.
Null pointer exception during interactive
installation of IBM HTTP Server on AIX systems
When installing IBM HTTP Server as a standalone GUI installation
on the AIX platform, you can receive a null pointer Java error on the installation
panels, when the installer begins to copy files to your machine. This situation
is caused by an unstable AIX Object Data Manager (ODM) registry.
To
work around this problem:
- Exit the installer.
- Issue the lslpp -l|grep IHS command to see
if there are remaining IBM HTTP Server registry entries that should not be
there.
- If there are remaining IBM HTTP Server registry entries, do a silent installation
of IBM HTTP Server by running the following command:
java -jar setup.jar -silent -P ihs.installLocation=<the desired install location>
Note: To remove the installation silently, add the -silent parameter
to the regular uninstall command, for example:
java -jar _uninst/uninstall.jar -silent
from
the IBM HTTP Server installation location.
Certificate Revocation List is not supported
for IBM HTTP Server on HP-UX and Linux/s390
Using Certificate Revocation List (CRL) with IBM HTTP Server HP-UX
and Linux/s390 is not supported at this time.
Problem using the ikeycmd command-line
interface from IBM HTTP Server on Linux/390
You can see a Java core dump after executing an ikeyman command
function, such as creating the stash file. This core dump occurs on both RedHat
and SuSE releases and is the result of a conflict in library routines caused
by the default loading sequence.
To work around this problem, set the LD_PRELOAD=/usr/lib/libstdc+//0-libc6.1-2.so.3 environment
variable before running the command. Then, this library loads first when the
application is starts. Note that setting this environment variable is also
necessary to allow Secure Sockets Layer to work correctly on Linux390.
Documentation
More information on configuring the Edge-Side
Include processor is located in the documentation article "Configuring Edge
Side Include caching"
ESI (Edge Side Include) is configured through
the plugin-cfg.xml file. This information is documented
under "Improving performance with the dynamic cache service". A link is missing
in the ESI documentation to the the plugin-cfg.xml file
documentation. The plugin-cfg.xml file documentation
is located in the article "Configuring Edge Side Include caching". Refer to
this article for more information on configuring the ESI processor.
Updates to documentation security article "Example:
Form Login"
This article contains an example of an error page in
a JSP file that does not work. Avoid using it in any application.
Using the View menu to increase the font
size
If section headings and formatted text do not work properly
when viewing the InfoCenter in your browser, use the View menu in your
browser to increase the font size.
Update for the article entitled "Cannot
uninstall an application or remove a node or application server"
The
article entitled, "Cannot uninstall an application or remove a node or application
server" contains a reference to a nonexistent addNode program option.
The
erroneous line reads:
"If the applications were installed indirectly
using the addNode program might be in use by other nodes. These applications
must be explicitly uninstalled, for example through the administrative console."
The
correct line should be:
"If the applications were installed indirectly
using the addNode program with the -includeapps option, then
removeNode does not uninstall them, since they might be in use by other nodes.
These applications must be explicitly uninstalled, for example through the
administrative console."
Update for the article "Creating a custom
Java administrative client program using IBM WebSphere Application Server
administrative Java APIs"
In the article entitled, "Creating a custom
Java administrative client program using IBM WebSphere Application Server
administrative Java APIs", replace the example under step 3 with the following
text:
@echo off
call "%~dp0setupCmdLine.bat"
"%JAVA_HOME%\bin\java" "%CLIENTSAS%"
"-Dwas.install.root=%WAS_HOME%"
"-Dwas.repository.root=%CONFIG_ROOT%" -Dcom.ibm.CORBA.BootstrapHost=%COMPUTERNAME% -classpath
"%WAS_CLASSPATH%;w:\DeploymentManager\classes;w:\DeploymentManager\lib\admin.jar;
w:\DeploymentManager\lib\wasjmx.jar" MyAdminClient %*
How to run a Performance Monitoring Infrastructure
application
The the article "Running your new monitoring applications"
shows a wrong script for using WSLauncher. The correct script follows:
call "%~dp0setupCmdLine.bat"
set WAS_CP=%WAS_HOME%\lib\properties
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\pmi.jar
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\pmiclient.jar
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\ras.jar
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\admin.jar
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\wasjmx.jar
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\j2ee.jar
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\soap.jar
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\soap-sec.jar
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\nls.jar
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\wsexception.jar
set WAS_CP=%WAS_CP%;%WAS_HOME%\lib\ws-config-common.jar
%JAVA_HOME%\bin\java "%CLIENTSOAP%" "%CLIENTSAS%" "-Dws.ext.dirs=%WAS_EXT_DIRS%"
%DEBUGOPTS% -classpath "%WAS_CP%" com.ibm.websphere.pmi.PmiClientTest host name [port] [connectorType]
Troubleshooting
All platforms
The Log Analyzer
displays error messages during startup
Upon starting the Log Analyzer,
the following error messages might display in the Log Analyzer shell window:
Cannot open input stream for default
Cannot open input stream for default
Cannot load configuration: default
Cannot open input stream for default
Cannot open input stream for default
Cannot load configuration: default
These error messages indicate corrupt
or incomplete user preference files.
To solve this problem, take the
following actions:
- Close the Log Analyzer.
- Delete all user preference files in the %USERPROFILE%\logbr directory
on Windows platforms or $HOME/logbr directory on UNIX platforms.
- Restart the Log Analyzer.
Note:Deleting all user preference files removes the preferences
of Log Analyzer set by the user in the preferences dialog.
The Log Analyzer
displays informational message during startup
Upon starting the
Log Analyzer for the first time or after the Log Analyzer preferences files
of the users are deleted, the following message displays in the Log Analyzer
shell window:
Cannot open input stream for waslogbrsys
This
message is an informational message. You can disregard the message because
it does not affect the execution of the Log Analyzer.
How to silently
install Log Analyzer
The responsefile.txt file for silent
installation needs more information to install Log Analyzer. To silently install
Log Analyzer, add the following option to theresponsefile.txt file:
-P logAnalyzerBean.active="true"
Set
the Performance and Analysis Tools property in the responsefile.txt file
to true for Log Analyzer to install. The property in the responsefile.txt file
follows:
-P performanceAndAnalysisToolsBean.active="true"
How to access
the Log Analyzer help files
For Windows platforms, you can only
access the Log Analyzer help files using the operating system default Internet
browser. You cannot access the help files using any Internet browser, even
though there are options allowing you to select either Netscape or Internet
Explorer and set the location of the browser to display HTML help files.
For
UNIX platforms, you can access the help files using any Internet browser,
such as Netscape Navigator, by explicitly setting the location of the browser
executable in the tool Preferences dialog. The option that seemingly allows
you to select either Netscape or Internet Explorer as the browser to display
HTML help files is not used.
To specify the browser on UNIX platforms:
- In the Log Analyzer tool, click File > Preferences.
- In the Log Analyzer Preferences dialog, click Help from
theGeneral folder.
- Set the path to the Internet browser executable in the Browser Location field.
Ignore the following
WebSphere Commerce Suite files
WebSphere Application Server includes
the following Log Analyzer files for use with the WebSphere Commerce Suite:
- install_root\bin:
- wcslogbr.bat
- wcslogbrsys.cfg
- wcslogbrsys.ini
- install_root\properties\logbr:
- wcsanalyzers.xml
- wcslogtypes.xml
- wcsrecdef.xml
You can ignore these files.
Errors occur setting trace specification for components
and groups
An error can occur when setting a trace specification
from the administrative console if selections are made from both the Groups and Components lists.
In certain cases, the selection made from one list is not lost when adding
a selection from the other list. To work around this problem, enter the desired
trace specification directly into the Trace Specification entry field.
Netscape browser fails when trying to enable
a component trace
On systems
using AIX, the Netscape browser fails when you try to enable trace on a component.
To
work around this problem, do one of the following:
- Disable JavaScript on the browser and continue setting trace.
- Administer the AIX server from a remote machine running another browser
and operating system.
- Change the trace manually in the server.xml file.
Enabling trace on client and stand-alone applications
The
WebSphere J2EE application client and many stand-alone processes define a
process-specific mechanism for enabling trace. To enable trace and message
logging for client applications or processes that have not defined such a
mechanism, add the -DtraceSettingsFile=filename system property
to the startup script or command.
For example, to trace the stand-alone
client application program named com.ibm.sample.MyClientProgram,
enter the command:
java -DtraceSettingsFile=MyTraceSettings.properties com.ibm.samples.MyClientProgram
The file identified by filename must be a properties
file placed in the classpath of the application client or stand-alone process.
An example file is provided in install_root/properties/TraceSettings.properties.
You can configure the MyTraceSettings.properties file to
send trace output to a file using the traceFileName property. Specify
one of two options:
- The fully qualified name of an output file. For example, traceFileName=c:\\MyTraceFile.log.
You must specify this property to generate visible output.
- stdout. When specified, output is written to System.out.
You can also specify a trace string for writing messages with
the Trace String property, Specify a startup trace specification
similar to that available on the server.
For your convenience, you
can enter multiple individual trace strings into the trace settings file,
one trace string per line.
Here are the results of using each optional
property setting:
- Specify a valid setting for the traceFileName property without
a trace string to write messages to the specified file or System.out only.
- Specify a trace string without a traceFileName property value
to generate no output.
- Specify both a valid traceFileName property and a trace string
to write both message and trace entries to the location specified in the traceFileName property.
