Oracle® Enterprise Manager

Hardware Requirements

The system or systems must meet, at a minimum, the following hardware requirements:

Software Requirements

The system or systems must meet, at a minimum, the following software requirements:

Install Issues

This section lists all the issues pertaining to installation.

Bug 4673479

For a shared cluster deployment, at the end of the deployment, the application will not bring up the deployment page. It will bring up the status page only. The status page shows details of deployment and you will be able to identify whether deployment succeeded or not.

Bug 4910745

Y ou cannot perform an additional 10.2.0.1.0 OMS install by using the repository of

10.2.0.2.0 OMS.

Workaround:

We can install additional OMS with the following workaround, if the repository is already upgraded to 10.2.0.2.0. To do this, perform the following steps:

1. Convert the repository version to 10.2.0.1.0 from 10.2.0.2.0 by using the following sql statement:

   UPDATE sysman.mgmt_versions SET version = '10.2.0.1.0' where component_name='CORE'; commit;

2. Provide the repository credentials for the Specify Database Configuration Screen and click Next.

3. Update the repository version to 10.2.0.2.0 by using the following sql statement:

   UPDATE sysman.mgmt_versions SET version = '10.2.0.2.0' where component_name='CORE'; commit;

4. Proceed with the installation.

Bug 5083308

For an agent installation in a Real Application Clusters (RAC) environment, the Oracle Configuration Manager (OCM) collector is only installed and configured on the local node. On the remote node, the OCM collector is installed but not configured. To configure the OCM collector on the remote agent nodes, construct an Oracle Enterprise Manager Grid Control OS job to complete the configuration. From the Oracle Enterprise Manager Grid Control home page, perform the following steps:

1. Go to Jobs Tab.

2. Select the host command from Create Job drop-down list and specify a suitable job name.

3. Go to Parameters.

4. Select command type to be os script.

5. Specify /bin/sh $ORACLE_HOME/ccr/bin/setupCCR -s <csi> <metalink-id> <country-code>.

Where <csi> is the Customer Support Identifier, <metalink-id> is the Oracle Metalink Id and <country-code> is the country code specified during the OUI install. If a proxy is required for the server to access the internet, the setupCCR command should also contain the –p qualifier specifying the proxy. $ORACHE_HOME/ccr/bin/setupCCR -h describes the command syntax for proxy specification.

Bug 5148669

If the OMS Patch Configuration fails while the EM deployment is in progress, the cause could be opmn proecesses are running from a different ORACLE_HOME, causing port conflicts. Please be sure all opmn processes in all ORACLE_HOMEs are stopped prior to performing the 10.2.0.2 upgrade.

Bug 5157782

If the sshUserSetup fails with "Bad owner or permissions on the file XYZ" error, then check the umask option on the local and remote machines by using the unmask command. Then, set the umask to 022 on the local machine before running the script. On remote machines, the umask should be set to 022 in the rc of the shell. For example, bashrc.

Bug 5158140

If 10.2.0.1 OMS is upgraded to 10.2.0.2.0 and if you want to deploy the 10.2.0.2.0 agents using the "Agent Deploy" application, then you need to manually stage the agent download kit into the OMS home. For more information on getting the agent download kit, refer to the "Enterprise Manager Grid Control Installation and Basic Configuration Guide".

Bug 5124872

The default platform agent download stage that gets installed with the OMS, will have the agent registration password in the response file that was used for securing the OMS, hence the default platform agent download installs can use this password from response file or overwrite it by setting AGENT_INSTALL_PASSWORD environment variable.

The agent download kits that will be manually staged into OMS home of a different platform of the agent download, the response file will not have the password, so if users want to secure the agent, they have to set the AGENT_INSTALL_PASSWORD environment variable and invoke the agentDownload script, else the agent download script will prompt for password.

Bug 5116075

When you deploy 10.2.0.1 agent using "push" method, in the host where agent will be deployed, the OS locale should be English locale (en_US). If the locale is other than English, such as zh_CN or ja_JP, it will fail in the pre-requisite check stage.

This issue has been fixed in 10.2.0.2 agent. To work around this issue, deploy 10.2.0.2 agent. If you need to deploy 10.2.0.1 agent, perform the deployment in English environment.

Bug 5037758

Cygwin on a different location, may be c:\ABCD\. But this has to be same on all of the remote hosts and OMS machine. To do this, you need to modify the ssPaths_msplats.properties file or the userPaths.properties file.

Bug 4954787

After an agent discovers a database (at the end stage of an agent installation) the status of database is "unavailable". The reason is that Enterprise Manager does not know the password of "dbsnmp" to monitor and manage the database.

Workaround

You need to logon EM, go to database list page and configure the database again.

Bug 4912661

In the Real Application Clusters (RAC) Agent install, after you reconfigure a secured Management Agent using the <AGENT_HOME>/bin/agentca -f script, check its status. Ideally, the Management Agent should have regained its secured status in spite of being reconfigured.However, if the Management Agent is not secured, you need to manually secure it by running the following script: <AGENT_HOME>/bin/emctl secure agent <PASSWORD>

Bug 4640583

When installing 10.2.0.1 Enterprise Manager into an existing database on a cluster that hosts an ASM database which is used for the repository, the following two issues arise:

The installer does not correctly support the use of ASM storage for the repository: To install the required Grid Control Tablespaces into ASM storage, you will have to manually determine the location for the files in the ASM directory structure. If ASM version 10.2.X is being used for the storage, the 'asmcmd -p' command can be invoked on the command line to determine the correct directory structure. The standard directory structure will look like "+DATA/<DB_NAME>/DATAFILES". Give a unique name to the tablespace datafile. Also, the suffix ".dbf" must be added to the data file.

The installer does not allow the installation of the OMS on the same node as an existing RAC ASM database that would used as the GC repository: To install Grid Control in an HA configuration to an existing RAC ASM database using the cluster nodes to host the OMS you must install the software on node 1 and point the OMS to the instance on node 2. A second OMS can be installed on node 1 by reversing the process and point the second OMS to node 1.

Bug 5171118

EM Grid Control needs read permissions on certain files under the Oracle Home of an installed 10.1.3 Oracle Application Server. This is needed both for discovery of the corresponding Application Server related targets in Grid Control and for monitoring of the Application Server.

This prevents Enterprise Manager Grid Control installed by a different user in the same Group as the iAS user from discovering and monitoring the above 10.1.3 iAS instance.

Workaround:

Install EM as the 10.1.2.0.2 or 10.1.3 AppServer user (until we get TBD patch releases for AS 10.1.2.0.2 and AS 10.1.3).

Bug 5086258 and 5084892

EM Grid Control needs read permissions on certain files under the Oracle Home of an installed 10.1.3 Oracle Application Server. This is needed both for discovery of the corresponding Application Server related targets in Grid Control and for monitoring of the Application Server. The out of the box install of AS 10.1.2.0.2 and AS 10.1.3.0.0 Application Server does not have the appropriate read permissions for the Group for these files.

This prevents Enterprise Manager Grid Control installed by a different user in the same Group as the iAS user from discovering and monitoring the above 10.1.3 iAS instance.

Workaround:

Install EM as the 10.1.2.0.2 or 10.1.3 AppServer user (until we get TBD patch releases for AS 10.1.2.0.2 and AS 10.1.3).

Bug 5178469

If you try to install an Agent from a non-default shiphome, you cannot use the agentdownload shiphome as it is. If you do so, the installation will fail.

To workaround this bug, perform the following steps:

1. Download the AgentDownload shiphome from OTN to some location, say /shared/mydir/, which is visible as /net/shared/mydir/ from target boxes.

2. Unzip it, you will get /shared/mydir/linux/ directory with the following contents:

   addons/

   agentDownload.linux

   agent_scripts.jar

   agent/

   oui/

   prereqs/

   response/

3. Rename or Copy /shared/mydir/linux/agent to /shared/mydir/linux/Disk1/

4. Use /net/shared/mydir/linux/ as the NFS shiphome location.

Bug 5069379

After upgrade, portlist.ini, setup.info, readme.txt, redirect.html may contain incorrect port assignments. Please keep continue using the same ports prior to upgrade as upgrade configuration migrates the old ports to the newer version.

Bug 5103331

To setup cygwin ssh server, you have to update the <cygwin dir>/etc/passwd file with the information about the user with which you want to connect to the ssh server.This requirement comes from cygwin sshd, not from agentpush.

Bug 5196643

The post install discovery script, agentca does not work with agent that is installed along with OMS. To discover any targets for the chained agent home, you have to manually discover them from the console.

Bug 5183333

When you install a 10.2.0.2 agent through Agent Deploy application using non-default shiphome option, it may fail with the following error:

SEVERE:Values for the following variables could not be obtained from the command line or response file(s): ToplevelComp(TopLevelComp) Silent install cannot continue.

Workaround:

1. Go to <oms-home>/sysman/agent_download/10.2.0.1.0 directory and open the agent_download.rsp file.

2. Replace the line TOPLEVEL_COMPONENT={"oracle.sysman.top.agent","10.2.0.1.0"} with TOPLEVEL_COMPONENT={"oracle.sysman.top.agent","10.2.0.2.0"}.

3. Save the agent_download.rsp file and try the installation again.

Bug 5211366

We cannot patch 10.2.0.1 agent to 10.2.0.2 using 10.1.0.x OMS, since patch cache size is insufficient to accommodate the 10.2.0.2 patchset.

Bug 5282471

During the deployment of a Management Agent using agent download method, the "sh: sysdef: not found" meesage is displayed.

Workaround : No functionality is affected due to this. However, set /usr/sbin in the PATH variable by executing export PATH=$PATH:/usr/sbin, and then deploy the Management Agent.

Bug 5498166

Aggregate Resource Usage Statistics (By Project) and Aggregate Resource Usage Statistics (By User) host metric display a "CIM error in method response" error.

Workaround : None. This issues is been tracked with bug #5498166.

Repository Issues

This section lists all the issues pertaining to repository.

Bug 5045731

If you are installing Grid Control on RAC environment and the repository owner schema already exists in the database, make sure that there are no sessions connected as repository owner. If you are creating Grid Control repository database using DBCA, uncheck the "Create database with Enterprise Manager" option.

Bug 4962360

When an Enterprise Manager repository is hosted on a 10.2.0.2.0 database, OMS connections to the repository are sometimes aborted with the following error message:

SQLException java.sql.SQLException: OALL8 is in an inconsistent state.

See Bug 4961570 for once such instance while managing Collabsuite targets.

Cause:

This is caused by RDBMS Bug 5033488 - ORA-07445: exception encountered: core dump [kokmrwo()+86].

Workaround:

Log on to the repository as sysdba and execute the following command:

SQL> alter system set "_optimizer_connect_by_cost_based" = false;

Solution

If the problem persists call Oracle Support.

Provisioning Issues

This section lists all the issues pertaining to provisioning.

Bug 5109061

The RefreshFromMetalink job runs every 24 hours, and pulls information on new product versions released by Oracle. This job needs to be run before being able to search Metalink for patches on 10.2 product versions. In case you would like to search Metalink for patches on 10.2 product versions within 24 hours of upgrading to EM 10.2.0.2, you should do a one time explicit run of the RefreshFromMetalink job.

Bug 5080749

In the Deployment Procedure Manager, some strings of Procedure tab, RAC Provisioning, and Patch Oracle Software screens are not translated.

Bug 4901328

This bug applies to the process of defining a Directive Step or a Component Step in Deployment Procedures. When setting values to properties that are supposed to be used as command line arguments, make sure to enclose the property value within quotes whenever the property value contains a white space character and you would want the property value to be treated as a single argument. As a rule of thumb, enter the property value with or without quotes, same as what you would normally enter through the shell command prompt.

Bug 5191216

Execution of a component or directive step from Deployment Procedure Manager fails if the staging area location or the directive file name contains spaces.

Workaround: Do not use spaces while specifying the staging location or directive filename.

Jobs Issues

This section lists all the issues pertaining to Jobs.

Bug 5028782

Suspending a job fails with "OALL8 is in an inconsistent state is logged in the emoms.log file." or if Patching Job fails with an initialization error "OALL8 is in an inconsistent state is logged in the emoms.log file."Cause:This is caused by RDBMS Bug 5033488 - ORA-07445: exception encountered: core dump [kokmrwo()+86].Related Bug:4961570Workaround:Log on to the repository as sysdba and execute the following:

SQL> alter system set "_optimizer_connect_by_cost_based" = false;Solution:If the problem persists contact Oracle Support.

Bug 5084633

A heavy load on Enterprise Manager Job System may cause some jobs to fail due to "initialization error". Also, the emoms.trc file shows the following exception at the same time:

eml.FxferRecv receiveFile.1629 - java.io.IOException: Too many open files

java.io.IOException: Too many open files

at java.io.UnixFileSystem.createFileExclusively(Native Method)

at java.io.File.checkAndCreate(File.java:1314)

at java.io.File.createTempFile(File.java:1402)

and

2006-04-10 14:03:52,636 [JobWorker 12478:Thread-136] INFO em.jobs executeCommand.526 - remoteOp: oracle.sysman.emSDK.emd.comm.CommException: IOException in sending Request :: Too many open files at oracle.sysman.emSDK.emd.comm.EMDClient.getStreamResponse_(EMDClient.java:3043) at oracle.sysman.emSDK.emd.comm.EMDClient.getStreamResponse_(EMDClient.java:2872)

Cause

This is caused by operating system running out of open file handles limit (default for linux is 1024)

Workaround

You need to increase the file descriptor soft limit per shell to 4096 from default 1024.

For Linux, perform the following steps:

1. # vi /etc/security/limits.conf

[Modify or add "nofile" (number of file) entries - note that a userid can be used in place of *]

* soft nofile 4096

* hard nofile 65535

2. # vi /etc/pam.d/login [Add the line]

session required /lib/security/pam_limits.so

For most systems this will be sufficient - log in as a regular user and check using command "limit" for tcsh and "ulimit" for bash shell. If limit does not increase, please check the operating system guide for increasing this limit.

Solution

If the problem persists contact Oracle Support.

Client System Analyzer Issues

This section lists all the issues pertaining to Client System Analyzer.

Bug 5095589

This bug affects the OS-registered software feature of CSA. In order to use this feature, the administrator must specify the full URL that the CSA applet will use to access the filter file. It is no longer sufficient to simply specify the file name."

Bug 5058225

There is no response metric for the CSA Collector Target in this release. This causes CSA collector targets to appear in the "status pending" state. However, the collector target will still function normally.

Agent Issues

This section lists all the issues pertaining to Agent.

Bug 5072520

Some files and executables are not accessible to the users who are in a different group than the group agents belong to.

Workaround:

To work around this problem, add the user to the same group as the agent user.

System Monitoring Plug-ins for Non-Oracle Middleware Issues

This section lists all the issues pertaining to System Monitoring Plug-ins for Oracle Middleware.

Bug 5086829

To enable the discovery and monitoring of IBM WebSphere Application Server targets on Solaris, you must apply one-off patch 5086829 (available on OracleMetaLink) to the Management Agent 10.2.0.2 that is to monitor the IBM WebSphere Application Server targets. For more information, refer to the patch's readme.

Bug 4451228

After the discovery of IBM WebSphere Application Server or BEA Weblogic Application Server, remove the entry $ORACLE_HOME/jlib/repository.jar from CLASSPATH entry in $ORACLE_HOME/sysman/config/emd.properties file and restart the Management Agent. This is needed only for the first time discovery of the Third Party Application Servers. You do not need to do this for subsequent discoveries.

Oracle Application Server Management Issues

This section lists all the issues pertaining to Metrics.

Bug 5012790

In order to see accurate values on the "mod_oc4j Metrics Page" for a version 10.1.3 Oracle HTTP Server, you must patch your Oracle Application Server 10.1.3 installation with patch 5161311 and patch 5088239 (available on Oracle MetaLink). Refer to the patches' readmes for additional information.

Bug 5004314

OHS to OC4J links do not appear on the topology pages for 10.1.3 Oracle Application Server and Oracle Application Server Cluster targets if the OMS and/or agent were upgraded to 10.2.0.2 from a previous version.

Workaround:

Execute the following command:

${OMS_HOME}/bin/emctl register oms targettype -refresh

${AGENT_HOME}/sysman/admin/metadata/oracle_apache.xml <repos-user>

<repos-pwd> <repos-host> <repos-port> <repos-sid>

Ensure that both the OMS and the agent have been upgraded to 10.2.0.2 before you run this command.

Bug 4904980

You cannot patch Ultrasearch at the instance level from Grid Control. However, you can still use Grid Control to apply Ultrasearch patches at the Oracle Home or host level.

Bug 4638457

The error dialog is garbled if starting a 'home' component fails. For example, the home component is disabled.

Workaround

Enable the component before you start it.

Bug 4093009

The CPU Utilization metric for OC4J instances on the Linux platform does not provide accurate data.

Bug 5195430

After disabling an Application Server component from ASC, you will still continue to see that disabled component from GC.

Workaround:

To workaround this issue, remove the disabled component from GC manually.

Oracle Collaboration Suite Management Issues

This section lists all the Oracle Collaboration Suite management issues in detail.

Bug 4638964

When starting the OCS components which are already with status "Up", there will be error messages which may be in the following three cases:

1) For Discussions, the error message is always consistent with OMS Server locale.

2) For OC4J_Mail, there are question marks in the error message.

3) For OC4J_RM, the error message is always consistent with CS Server locale.

Workaround

Start both OMS and CS servers in pure English environment.

Bug 4670523

The translated Japanese Online Help for the Mobile Collaboration metrics on the Mobile Collaboration All Metrics Page is not currently available.

Bug 5172122

If the system and service name contains multibyte characters the confirm cancel from intermediate page of OCS systems and services wizard now comes back to ocs tab page.

Bug 4961570

If Enterprise Manager repository is hosted on a 10.2.0.2 RDBMS and you use Enterprise Manager to manage Collabsuite targets, you could encounter the following bug. Enterprise Manager console reports an internal system error while creating an IM service:

SQLException java.sql.SQLException: OALL8 is in an inconsistent state is logged in the emoms.log file.

Cause:

This is caused by RDBMS Bug 5033488 - ORA-07445: exception encountered: core dump [kokmrwo()+86].

Workaround:

Log on to the repository as sysdba and execute the following:

SQL> alter system set "_optimizer_connect_by_cost_based" = false;

Solution:

If the problem persists contact Oracle Support.

Solaris Issues

This section lists all the issues relating to Solaris OS.

Other Issues

This section lists all the other issues.

Bug 5097695

Generic Log File Monitoring functionality [available as part of host target functionality] requires correction to the Solaris host metadata file. Associated post installation tasks are documented in the Post Installation section of this document.

Bug 4922931

When you set a report's time period, the custom date format doesn't support the NLS native date format. So, it fails with invalid date format error because it is not UTF-8 encoded. The workaround is to use english as browser language only when you set a time period.

Bug 4880147

An automatic restart option can be selected for the Fast-Start Failover observer when configuring Fast-Start Failover from Enterprise Manager Grid Control. If there is a subsequent role change (either switchover or automatic failover) in the Data Guard configuration while Fast-Start Failover is enabled, the observer restart feature will be disabled. It can be re-enabled by re-running the Fast-Start Failover configuration process, as follows:

1. On the Data Guard overview page for the configuration, click the Enabled link to re-run the Fast-Start Failover setup wizard.

2. Select the "Attempt automatic restart of the observer" checkbox and click Next.

3. Proceed through all subsequent screens and click Yes on the confirmation screen.

After the configuration process completes, re-run the Fast-Start Failover wizard again, select Edit Properties, and ensure the "Attempt automatic restart of the observer" checkbox is checked. This will verify that the restart option is enabled.

Bug 4770372

An error is generated in the Grid control TAB Management services and Repository everyday at 12.00 AM.

Workaround:

Logon to SQL as repository owner and execute the following:

SQL> exec em_purge.disable_purge_policy('MGMT_POLICY_ERRORS') ; commit ;

Bug 4714082

Grid Control hangs while getting database connection to the repository.

Workaround:

Killing and restarting the tnslsnr processes on the repository resolves the issue.

Solution:

Contact Oracle Support to obtain the patch for RDBMS Bug-4518443.

Bug 4655690

After an upgrade, the performance and usage graph on the Service home page will be missing. You will need to go to "Monitoring Configuration", then to "Performance Metrics" to add the metric and select the chart to be displayed on the home page. You can follow similar steps for "Usage Metrics".

Bug 5151931

When Grid Control is using a 9.2.0.6 or 9.2.0.7 remote DB, you need to apply BLR for RDBMS (Bug 4523371) from ARU.

Bug 4652602

The Licensing Model for Web Applications is slightly different as explained below.A Web Application target has SLM and all AS related packs (as_diag(AD), as_config(AC), Provisioning(PV)) associated with it. All Web Applications are "automatically granted license" on SLM pack, when the Web App is created or when a Web App that exists in 10GR1 is upgraded to 10GR2.The Web App cannot be granted license on AD and AC packs directly. The Web App instead derives licenses on these packs from AS and AS Dependent Targets (OC4J, LDAP etc) that have these packs licensed, it contains. In case the WebApp that existed in 10GR1 did not contain an AS or AS Dependent target as a key component, then this WebApp when upgraded to 10GR2 is not licensed for AD and AC packs.Hence the links this bug talks about, which need this pack licensed are disabled. This is normal expected behavior. The "Setup | Management Pack Access | All Targets" UI allows users to grant licenses to targets on any packs applicable to them, bypassing business rules. This UI lists all four packs (SLM, AD, AC, PV) for a Web App Target and you can grant/revoke licenses on these packs to the Web App target.On granting license on desired packs to WebApp Target, the above links will be enabled.

Bug 4614984

Multibyte webapp name does not display correctly in Root Cause Analysis page. All three fields in RCA page "Step", "Target Name" and "Message" will display multibyte string as garbage characters and square boxes.

Bug 4999766 and 5150067

The Interactive Transaction Trace and Request Performance functionalities are not supported for Web Applications running on Oracle Application Server release 10.1.3.0. A fix for this problem will be available with Oracle Application Server release 10.1.3.1. and GC 10.2.0.3.

Bug 5174330

To workaround this bug, perform the following steps:

• After Refresh From Metalink job is run(scheduled or manually), refresh the browser if it was active before the job was started.

• After Refresh From Metalink job is run(scheduled or manually), users can re-login to Enterprise manager if they were logged in before the job started.

• After Refresh From Metalink job is run(scheduled or manually), users can go to CPF main page by clicking on "Critical Patch Advisories for Oracle Homes" link on Enterprise Manager Home page. From CPF main page, users can click on "Refresh Data" button provided on top right side of the screen. This will restore the CPF session data.

Bug 4517767

When you are backing up iAS mid-tier installation types of 9.0.4.x versions, you may encounter the following error:

Could not copy file /scratch/904/Apache/Apache/conf/osso/osso.conf to /scratch/904/conf1/2005-08-09_06-52/Apache/Apache/conf/osso/: No such file or directory.

To resolve this, follow the steps described in the iAS Release Notes and retry the backup process.

Bug 4407853

This applies if your envirnonment is using an Enterprise Manager 10.1 OMS and a 10.2 Agent (either upgraded from version 10.1 or a new install).

When an agent is upgraded to version 10.2, the Web Application targets on that agent will be upgraded to type_meta_ver 3.0. The version 10.1 OMS does not have Web Application Response metric for type_meta_ver 3.0. This will cause problems with the computation of Web Application status. A patch must be applied on the repository to create the Response metric metadata for type_meta_ver 3.0. This patch is required to show the correct Web Application target status.

Bug 5073502

On Oracle Clusterware / RAC Provisioning, Install Inputs page and Cluster Database name should not allow Multibyte chars. At present, no validation is done for that.

Bug 5012484

SGA Functionality is not supported in 10.2.0.2.0 GC.

Bug 5194444

In only Brazilian Portuguese's Job View, the list of values which are Runs and Executions are translated into the same word. As a workaround to distinguish them, you can change the browser language as english or download the one-off patch of bug5194444 from Oracle MetaLink.

Bug 4918883

When Grid Control bounces at fr_FR.utf8 (or other western-european language based locale in which a dot(.) is displayed as a comma), and if you click the number link on the dashboard page for the OCS Service, "Internal Server Error" will show up. This is a NLS specific issue and it caused by incorrect number parse. This bug will be fixed in GC 10.2.0.3 patchset.