Symptom descriptions and detailed workarounds for troubleshooting when you are using the Agent Builder or when you are working with your agent.
See the Tivoli Monitoring: Troubleshooting Guide, GC32-9458 for general problem determination. Also, see the IBM Tivoli Monitoring Messages Guide for general IBM® Tivoli® Messages.
(Table 1) shows solutions for installation, configuration, and uninstallation problems.
Problem | Solution |
---|---|
Maximum data size exceeded When you save an agent,
you might see the following error in the Problems View:
|
Tivoli Monitoring currently limits the amount of data that can be returned
for each row in a table to 8192 bytes. For each data source in the
agent, the Agent Builder computes the sizes of the attributes in the
following ways:
|
An error occurs when you install Tivoli Enterprise Monitoring support on Windows system I tried to install Tivoli Enterprise Monitoring support on a Windows Tivoli Enterprise Monitoring with an Agent Builder script, and
received the following error:
|
Reset your PATH to ensure that the Windows find command is found first. It is possible that something was installed that is now the first find command in the PATH. For example, if you installed Cygwin (a set of UNIX tools for Windows). If Cygwin is first in your PATH, you hit the Cygwin find command rather than the built-in Windows find command. |
Display name is not refreshed If you are browsing to add a process or a service and select one, and then you select a different one before you click OK, the display name is not refreshed for the new process or service that is selected. |
Delete the characters that are in the display name field, and then browse. |
Demo CD problems I am running the demo CD as my Tivoli Monitoring system. I selected the managed node to browse. I am shown an unusual set of processes and UNIX operating systems is automatically selected for my process instead of Windows. |
You selected the wrong agent. The Demo CD contains many agents which would not normally be found together on a system. Stop the other OS agents and browse by using the real Monitoring Agent for Windows. Alternatively browse by using a system that has a normal Tivoli Monitoring installation. |
The -console flag does not work You receive a message:
You try to run the installer with the -console flag, but it does not work. |
The silent installation does not support the -console option. It does, however, support the -silent option. For more information, see (Silent installation). |
No help is displayed when you click help links within the Agent Builder On an AIX system, you might see the following messages: The context help for this user interface element could not be found. Open in dynamic help. Clicking the Open in dynamic help link does not do anything. |
On some AIX and Linux systems, Eclipse is unable to determine the default system browser. To correct this issue, you must set the default browser in Agent Builder. For more information, see Set the default browser in Agent Builder. |
Agent Builder does not start on a 64-bit Linux system Agent Builder does not start on my 64-bit Linux system. No errors are displayed or logged. |
Make sure that you have the prerequisite software installed. Agent Builder is a 32-bit application so the 32-bit version of the prerequisites must be installed. For more information, see Requirements |
Linux Gtk library warning messages Even
after I install the latest Linux 32-bit libraries for Agent Builder I see the following messages:
|
These messages can be ignored. The pk-gtk-module and canberra-gtk-module packages do not have to be installed. |
Problems running the Agent Builder on a non-English system When you run the Agent Builder on a non-English system, the JVM might have difficulty to determine the locale of the system. This issue prevents the Agent Builder from running. |
To get around this issue, edit install_location/agentbuilder.ini and
add the following 2 lines:
Where LOCALE_PREFIX is the
prefix for the locale the system is running (that is, FR for French).
Also, the -vm line and the line immediately
after must remain the first two lines in the file. Add this option
AFTER those lines but before the -vmargsline:
|
Remote service browsing error A failure occurs when you enter the host name and IP address of the local system when you try to browse WMI or services remotely. |
Browsing in this way is not supported. You receive an error that the builder was unable to open a connection to the host. If you want to browse WMI or services locally, select localhost from the list instead of entering the host IP address. |
Key attribute is missing When you save an agent, you might
see the following error in the Problems View:
|
You must have key attributes in the following
situation only: If the data source can return multiple rows of data, select an appropriate attribute or attributes as keys. If the data source can return only one row of data, select Produces a single data row for the data source. There are other side effects too. Situations can combine data from multiple attribute groups which are defined to return only one row. |
Change command return code or script data provider executable file When you change a command return code or a script data provider executable file, newly defined return codes do not show up. |
If you change the executable file, then you must copy the new executable file into the agent's project directory under the scripts subdirectory. Copying ensures that the new executable file is picked up when you regenerate the agent. |
Error when you attempt to start Eclipse: Agent Builder fails to start and displays an error that points to a log file in the configuration directory. The log file contains: No application id has been Found. |
Uninstall Agent Builder, and reinstall it in a fresh directory. |
Browsing processes or services on a remote system do not work |
If use the browse remote system function, the Tivoli Enterprise Monitoring Server, the Tivoli Enterprise Portal Server and the OS agent are required to be running. Restarting the Tivoli Enterprise Monitoring Server or starting it after the OS agent causes the OS agent to go offline. The agent goes offline until its heartbeat with the Tivoli Enterprise Monitoring Server is done. Alternatively the OS agent can be restarted. |
When I generate an agent and omit the .bat or .cmd in the script data source command, the script does not run. |
Scripts in Windows are frequently started without specifying the .bat or .cmd extension on the command line. Do not omit the .bat or .cmd in the script data source command. |
Resource out of sync I edited my agent xml file outside the project. I then get this error
when I open the project:
|
To fix the error, click the project and press F5 to refresh. Then, close the Agent Editor and reopen it. |
Parsing MIBs When I load my MIB into the Agent Builder, it did not parse. I see is something like the following messages:
, and
other similar messagesMibParser.java:1198 and MibParser.java:1081 might point to other areas of the parser. |
To get the MIB to load: The MIB parser that is used by the Agent Builder uses the grammar
that is defined by ASN.1 to parse the MIBs. Some MIBs do not follow
the grammar correctly. The parser can relax certain rules to accommodate
the most common errors. The MIB must be corrected, but until that
can be done you can turn off checking for common errors by doing the
following steps:
|
MIBs frequently have errors |
As many of these errors as possible are fixed by the Agent Builder. However, some errors are so severe that they cannot be corrected. In many cases, it is possible to edit the MIB, correct the file, and then attempt to import the MIB into the Agent Builder. As MIB errors are discovered, IBM, customers, and IBM Business Partners add information to the A&BSM Technical Exchange Wiki, IBM Tivoli AA&BSM Technical Exchange. The information indicates the MIBs on which errors were discovered and the corrections that can be imported into the Agent Builder. See Parsing MIBs earlier in this table. |
Agent Builder version on UNIX The Agent Builder is installed on a UNIX system, and you cannot start the GUI to see the version. |
The directory name that
matches the following pattern contains the current version of Agent
Builder:
or
|
Adding new attributes to log file data source Next field is by definition a way to build a log record from a log file. The record is built in the order in which fields are encountered in the line that is being parsed. IBM Tivoli Monitoring requires that all new attributes be added to the end of an attribute group. |
Generally, application logs are what they are
and when they change they can change dramatically. It is also often necessary to support the new and old log file formats to correctly support current and old versions of the application. The best way to handle this issue in an agent is to add an attribute group that represents the new log file format. When the agent is deployed to a system, you normally configure the name of the file to be monitored in the appropriate configuration variable. Use the configuration variable so that only the old or new attribute group is used in a specific instance of the agent. |
Agent Builder does not allow duplicate data source names in different subnodes In the portal Navigator Physical view, I want the same Navigator node name for data sources that are in different subnode types. |
The Agent Builder prevents you from adding warehouse
data from two different attribute groups that have the same name in
the same agent. A possible workaround to get duplicate names in the Tivoli Enterprise Portal Navigator is to define Navigator groups in the subnode definition. You define in the subnode by using the name that you want to display in the portal. Move the data source into the Navigator group and the data source name is effectively hidden. |
CIM integer array value Only the first value of my CIM Integer Array was returned. |
The Agent Builder currently does not have an
array data type. String Arrays can contain variable length data, all
of the strings are concatenated into a single value that is separated
by a comma (,). Verify that the overall length of the attribute is large enough to contain all of the expected values. The Integer data type attribute does not accept the, delimiter. Rather than convert the integer array to string data, only the first value in the integer array is returned. |
No CIM provider support for certificates No information is displayed in the Tivoli Enterprise Portal. The Agent Builder CIM provider does not support the use of certificates. |
CIM provider support is provided only for base, plain HTTP, and HTTPS. The CIM server was configured to require SSL certificates for all communication. The only workaround is to reconfigure the CIM server to allow communication without the use of SSL certificates. |
How do I display large numeric values? The browsers create all of my numeric attributes as signed 32-bit values when compatibility with IBM Tivoli Monitoring V6.2 is required. These 32-bit values do not work for unsigned data or large values that I want to collect. |
See Negative or wrong attribute value for a technique to use derived attributes to display large numeric values by using signed 32-bit attributes. |
Parse Log window characters corrupted Characters that are displayed in the Results area of the Parse Log window are corrupted. I expected to see my non-ASCII file contents, displayed correctly in the browser and in the portal after I deploy my agent. |
The Agent Builder log file parsing does not support parsing files with names or contents that contain non-ASCII characters. |
Browse not active when you add a data source When you add a data source, Browse is not active when the data source type is WMI, Perfmon, or Windows Event Log; and Agent Builder is running on Linux or AIX. |
For the Agent Builder on Linux or AIX, you cannot use the Browse function when you are creating WMI, Perfmon,
or Windows Event Log data
sources. The Windows APIs
must be available on the Agent Builder system to Browse these data
sources. You can still define these types of data sources when you run the Agent Builder on Linux or AIX by specifying the data source information manually. |
Agent Builder CIM Browser does not show CIM Classes for AIX OpenPegasus 2.6.1. Efix 644427.080123 for sysmgt.pegasus.cimserver.rte - Fix for PAM stack overflow Vulnerability in OpenPegasus 2.6.1 for AIX is applied to the IBM Pegasus CIM Server. |
This issue is resolved in IBM Pegasus CIM Server V2.6.1.35 You can display the current version of the IBM Pegasus CIM Server file sets by using the following file: lslpp -l sysmgt.pegasus.cimserver.rte Upgrade to IBM Pegasus CIM Server to 2.6.1.35 or later For current details, see the AIX Common Information Model overview OR Temporarily remove EFIX 644427 to browse the IBM Pegasus CIM Server on a development
CIM Server and allow browsing of the CIMOM.
Note: Be sure that you
have access to the EFIX 644427.080123.epkg.Z file
to reapply the EFIX.
Remove the EFIX:
(continued on the next page) |
Agent Builder CIM Browser does not show CIM Classes for AIX OpenPegasus 2.6.1. (continued) |
Reapply the EFIX:
|
Agent Builder CIM Browser does not show CIM classes for Solaris WBEM Server. The Agent Builder CIM Browser connects to a Solaris WBEM Server, but when a Namespace is selected, the CIM Browser displays the following error: KQZ0224E An unknown error occurred when attempting to connect to the CIM server on host hostname. |
This issue is reported to SUN and is a Vendor
Limitation. There is an error in the Solaris WBEM CIMOM. The enumerateClasses
method is logging a NullPointerException in the WBEM log. There is no workaround. |
JDBC connection to z/OS® DB2 database is failing. |
Make sure that you are using the correct set of JAR files. The z/OS DB2 connection fails if you are not using the JAR files that include the correct licensing information. |
Specifying the internal_logon connection property for Oracle How do I specify the internal_logon connection property for Oracle? |
If you want to allow users to specify the internal_logon
connection property, add the following configuration property to the
agent XML file. Add this property to the runtime configuration section
of the XML file after you create and saved the agent with at least
1 JDBC data source. Find the KQZ_JDBC_PASSWORD property and insert
the following text after that property:
|
The Agent Builder uninstallation ends with an error. If the Agent Builder is running when you attempt to uninstall, the uninstaller ends with an error. |
Close the Agent Builder and manually delete the Agent Builder installation directory to complete the uninstallation. |
Project default location error On the New IBM Tivoli Monitoring Agent Project page, if you clear the Use default location check box, and then browse to, or type, the default location, Eclipse gives an error that the specified directory overlaps an existing workspace. If you type a subdirectory underneath, Eclipse gives an error that the directory overlaps an existing project. |
If you want to use the default location, select the Use default location check box. Do not browse to, or type, the default location. |
Product code and affinity errors not cleared In the Agent Builder Editor, if you modify the product code or affinity (company ID or agent ID) of an agent, Agent A, so that it overlaps with the product code or affinity of another agent, Agent B, in the workspace, an error is created against Agent A. If you then modify the product code or affinity of Agent B so that the product codes or affinities are now unique, the error still exists on Agent A. |
|
Agent Builder script argument parameterization does not work. When you create a script agent, passing a runtime Parameter as an argument to the script at execution time does not work. |
The command line cannot include
environment variables. Runtime parameters must be referenced inside
the script. The command cannot be script ${K51_PROCNAME} Instead, inside the script, reference the variable:
|
An error message in any of the Agent Builder wizards is truncated. |
Resize the window to make it wider so that the entire message can be viewed. |
JRE or JDK not available You receive the following message: A Java Runtime Environment (JRE) or Java Development Kit (JDK) must be available in order to run Agentbuilder. No Java virtual machine was found after searching the following locations: C:\Program Files\IBM\ITM\AgentBuilder/jre/jre/bin/javaw.exe. |
Open the install_dir/agentbuilder.ini file and modify any lines that have a leading space by removing the leading space. |
Installing the agent locally fails with error message KQZ0208E When you generate the agent, install it locally and specify a password with special characters, the login to the monitoring server fails with error message: KQZ0208E The specified user name or password is incorrect. |
The password cannot include special characters. |
iFixes applied to the Agent Builder are not removed when the Agent Builder is uninstalled If you apply an interim fix to the Agent Builder, when you uninstall the Agent Builder, the interim fix is not removed. When uninstalling, an installer message informs you that it was unable to remove all directories. |
Remove the interim fix directories manually before you reinstall the Agent Builder. |
Agent Builder workspace screen not responding in UNIX over a Cygwin Xserver The Agent Builder workspace screen does not respond in UNIX over a Cygwin Xserver. When you start Agent Builder, the options on the workspace screen are unavailable and you are unable to do actions. |
An Eclipse issue with the Xming and Cygwin tools and not an issue with the Agent Builder. There is no fix available. |
Using the Parse function on the Parse Log window, returns incorrect results, such as no rows shown When you use the Parse function on the Parse Log window, incorrect results are returned, such as no rows shown. |
The Test log file parser reads only the last 1,000 lines of the log file in tail mode. Make sure the data that you are trying to test for occurs in the last 1000 lines of your sample log file. |
JMX browser fails to connect to WebLogic 10.x The JMX browser fails to connect to WebLogic 10.x. The configuration settings are correct, but connection errors in the traceKQZ.log file show that a CORBA.MARSHAL error occurred. |
Copy the wlclient.jar and the wljmxclient.jar files from your WebLogic server installation into the jre/jre/lib/ext directory of the Agent Builder installation and restart Agent Builder. This action loads the WebLogic JAR files in the system class loader which makes sure the correct classes are loaded to create the JMX connection. After you create your attribute groups by using the JMX browser, you must remove these two JAR files and restart the Agent Builder. |
When I click Collect Data in a Test window, no data is returned or the data is not what I expect. | Examples: Log file provider might not detect and process the file yet. Ping might report all zeros since the ping did not happen yet Click Collect Data a second time If this solution does not resolve the issue, you can click Check Results. The Data Collection Status window opens and shows you more information about the data collected. The data that is collected and shown by the Data Collection Status window is described in Performance Object Status node You can attempt further debugging by looking at the test log files, see (Figure 2). |
Problem | Solution |
---|---|
Program fails No data
for a script provider or a command return code is unusable. You see
the following in the trace log:
This shows that it tried to run
the command script1 with the argument not.bat. |
If you call a program with spaces in the
name, use quotation marks around the name so that it is not parsed
by the command interpreter. For example, this is a test.bat
argument becomes:
|
Negative or wrong attribute value I created an agent that includes a number that I expect to be a large positive number, but I see a negative or incorrect number. |
Tivoli Monitoring 6.2 uses 32-bit signed integers to represent numeric values. A 32-bit signed integer can display values from -2,147,483,648 to 2,147,483,647. The value that you are trying to display overflowed the 32-bit signed number. In many cases, the values are traced to the log file and replaced with enumerations that indicate the value exceeds maximum or minimum. Overflows can usually be handled by creating another attribute that scales the large one to a more reasonable value. For example, if the number represents
the size of the disk in bytes, it is more useful to use megabytes
or even gigabytes. Use the following procedure to convert the value:
You can now either leave the original attribute or hide it by selecting the attribute and clearing Display attribute in the Tivoli Enterprise Portal. Hide the original attribute if it is likely that it might overflow a 32-bit signed integer. Tivoli Monitoring V6.2.1 introduces 64-bit numeric attributes. Changing a 32-bit numeric attribute that is overflowing to a 64-bit value is a natural way to represent large numeric values. |
JMX Notifications I want create a data source to receive notifications, but I cannot see an option to configure in this way. |
When you use the browser, take note of the MBean
information that is displayed in the browser panel for the MBean you
are working with. If the MBean contains items in the Attributes tab,
then you get an attribute group that contains those metrics. If the
MBean contains items in the Notifications tab, you get an event attribute
group that contains a standard set of metrics for a Notification object.
If both tabs contain values, then you get attribute groups for each. If notifications are not defined for the MBean, the browser does not create the event attribute group. You can create the event attribute group manually by not clicking Browse to display the JMX browser. Instead, manually type in the Object Name pattern and click Finish. This action creates the same event attribute group that you get from the browser had notifications been detected. It also creates an attribute group to receive data with no attributes defined. This other attribute group can be deleted. |
Locally configuring an agent fails on Windows An Agent Builder created agent that contains configuration properties requires Tivoli Monitoring to install the Java Runtime Environment. If the Monitoring Agent for Windows OS was installed by using the tacmd createnode command to create a Tivoli Monitoring v6.2 Fix Pack 1: Windows OS Agent, then the Java Runtime eEnvironment is not installed. On this system, you cannot locally configure any Agent Builder created agents that contain configuration properties. |
The agent can still be configured remotely by
using the Tivoli Enterprise
Portal. The following solutions can be used to install the Java Runtime Environment that is
needed by Tivoli Monitoring:
|
Command line does not allow me to configure an agent I cannot use the itmcmd config command (or CandleConfig) to configure an agent that contains JMX, SNMP, or JDBC attribute groups. |
You must have Tivoli Monitoring v6.2 Fix Pack 1 to configure an agent that contains JMX, SNMP, or JDBC attribute groups by using the itmcmd config (or CandleConfig) commands. You can use the GUI to configure the agent if an upgrade is not an option. |
JMX monitors JMX Monitors are not working with the JBoss application server. |
The following is needed to make JMX Monitors work. Copy the connJboss-1.0.jar file from CANDLE_HOME/TMAITM6/kxx/jars/common/connectors/jboss on Windows or CANDLE_HOME/dynarch/kxx/jars/common/connectors/jboss on UNIX to JBoss_install/server/default/lib. If you have a server configured other than the default server, the default part of the path is different for your server. If the JBoss server is running, it must be restarted after you copy this file. |
JMX browser cannot connect to WebSphere 6.1 The JMX MBean browser cannot connect to WebSphere 6.1 with security enabled, when you use the SOAP connector protocol. |
The browser can collect the MBean names by using the rmi connector protocol. |
Error code for attribute group, but data is being returned The Performance Object status error code for my JMX attribute group is ATTRIBUTE_ERROR. Data is being returned for the MBeans, so what does this mean? |
A JMX attribute group that has an error code
set to ATTRIBUTE_ERROR in the performance object
status attribute group has one or more attributes that cannot be collected.
This error not only indicates a problem with one or more attributes,
but it is a performance issue as well. To determine which attributes are having a problem, look for the exceptions in the JMX trace log file. The exceptions typically indicate the class path cannot locate a certain class or the attribute object cannot be serialized. When you see this error, the attributes must be collected individually from the MBean Server instead of collecting all attributes in one remote call. This collection method can significantly affect the performance of the agent. |
JMX data provider cannot connect to WebSphere 7.0 JMX data provider fails to connect to WebSphere 7.0 with security enabled, when you use the RMI connector. |
An issue with the WebSphere Application Server. Upgrade to WebSphere 7.0.0.1 or later to resolve this connection issue. |
JMX data provider cannot connect to Oracle WebLogic Server on AIX The JMX data provider fails to connect to the Oracle WebLogic Server when you run on AIX. |
This connection problem is a limitation that is documented by Oracle WebLogic. Connect to Oracle WebLogic remotely from an operating system with a compatible Sun Java runtime environment. In this case, the agent must be configured to use the compatible Sun Java runtime environment instead of the IBM Java runtime environment. |
Title bar not displaying the workspace name When you look at a generated agent through the Tivoli Enterprise Portal, the title bar displays Nav_Node - TEPS_Hostname - UserID rather than Workspace_Name - TEPS_Hostname - UserID. |
When an agent is generated, workspaces are not created by default. If no workspace is defined for an attribute group, the Tivoli Enterprise Portal displays a generated workspace for the attribute group that shows all attributes in a table view. The title bar of the Tivoli Enterprise Portal displays the Navigation Node ID because the workspace name is not defined. You can click Creating workspaces. to save the generated workspace as the default workspace. A workspace name is now displayed in the title bar. For more information, see |
The log file is not monitored properly If you build an agent with a log file data source, and all or part of the log file name comes from a Configuration Property, and that Configuration Property value contains a space, the log file is not monitored properly. |
Enter the Windows short name for the file or path in the agent configuration. You can get the short name of any file or directory with the DIR /X command issued at a Windows command prompt. |
Agents not translated Agents that are created by Agent Builder are not translated when shown on the Tivoli Enterprise Portal. |
The Agent Builder does not build language packs. This means that the text that is displayed in the Tivoli Enterprise Portal is in the language you used when you built the agent. |
RAS1 log errors I see the
following in the RAS1 log for my agent. What does it mean?
|
This error can occur on some systems for the process that represents the system in Windows. The process environment block is not available for this process. System is a special process and normally is not one you are monitoring. If all of the data in your Availability table is completed, then this error does not represent a problem. You can verify by checking the PID printed in the trace against the PIDs for the processes you are monitoring. |
Log file data mismatch A record from the log file is not displayed on the Tivoli Enterprise Portal. Or the last entry or last few entries on the Tivoli Enterprise Portal do not match the contents of the log record. |
Check for error messages in the trace file,
for example, HOSTNAME_81_k81agent_465c087e-01.log. Look for the following trace entries: (465C08D7.0000-2A4:logmonitorqueryclass.cpp,506,"LogMonitorQuery |
Installed agent does not show up An installed agent does not show up in the Tivoli Enterprise Monitoring Services utility. |
Select View > Refresh from the Manage Tivoli Enterprise Monitoring Services window. |
Situations not showing up I expect the situations to be true but they are not displayed as true. Or, I expect my node to display the little red circle that indicates that situations are true. I checked and the data exceeded the threshold. Why do I not see the situation on my node in the navigation tree. |
Some of the associations between nodes and situations are loaded when the Tivoli Enterprise Portal starts. Restart the Tivoli Enterprise Portal. |
Queries not showing up The new queries are not showing up. |
Install the agent and then recycle the Tivoli Enterprise Monitoring Server and Tivoli Enterprise Portal. |
Error states the startagent command failed An error on the Tivoli Enterprise Portal says that the C:\IBM\ITM\InstallITM\Batch\kincli -startagent -akxx command failed. |
After you install Tivoli Monitoring on a Windows endpoint, the machine must be rebooted before it can successfully be a target of remote deployment, or the start, stop, and remove functions that are available on the Tivoli Enterprise Portal. The reason for this is that the Tivoli Monitoring installation changes the PATH of the system to include some DLLs that must be found for those functions to work. However, the services do not pick up those changes until the machine reboots. |
When you run command return code, the return code is not accurately shown When you run the Windows command return code to run a command and analyze the return code, and the command is a .bat or .cmd script, the return code is not accurately shown. The script exits with something like exit /b 2. |
To get this value, surround the target script with a script that calls the target script and then runs"exit %ERRORLEVEL%". |
Attribute group data is missing from Tivoli Enterprise Portal I built an attribute group by using a script source and I do not see any data in my Tivoli Enterprise Portal. |
Read the log file and look for text that
looks like:
This text indicates that your script returned data that did not match the defined format (in this case, items that are separated by a colon). It was attempting to parse the string that is contained within the first <> pair. Fix the script to return data in the correct format. You can test the result by editing the script in the agent directory because it calls the script each time it tries to collect data. |
Changing version number causes errors If you have an agent that you created with the Agent Builder, and you modify it and change the version number, if you attempt to deploy the agent remotely with the new version, you get an error that says KFWITM291E An agent configuration schema was not found. |
The Tivoli Enterprise Monitoring Server and Tivoli Enterprise Portal Server support must be installed again, even if you changed nothing but the version number. This error is caused by certain configuration files that contain the version number. You receive failures if the files with the new version number are not present. |
Changing product code, company identifier, or agent identifier causes problems You have an agent that you created with the Agent Builder, and you modify it and change the product code, company identifier, or agent identifier after you created workspaces or situations. |
All the situations and workspaces must be re-created. |
A script provider behaves oddly when edited For instance, you edit the script and add a sleep 30 to simulate a timeout. The timeout occurred as expected during the next refresh of the group. Then you take out the sleep 30 and refresh the group again. The calculated values are now all set to 0. |
This behavior is because of the previous data point being lost. |
Core memory dump after you edit the .ref file Core memory dump with a seemingly innocent modification to the .ref file. For example, splitting attributes of an element in the .ref file onto different lines still produces valid XML, but the agent coredumps. |
Do not edit this file. |
The Tivoli Enterprise Portal does not show any columns or column headers Navigator groups in an agent do not show any columns or column headers. Instead, an error is displayed in the view: KFWITM220E Request failed during execution. In addition to navigator groups that the user created, this error is also seen with internally generated navigator groups such as the Availability and JMX Monitors navigator groups. |
When you define a navigator item, there can be more than one attribute group with a node in the navigation tree. When there is more than one attribute group, you must assign a query to a workspace. After you assign a query, you can see the data. |
Bad string values that are collected from SNMP are displayed in attribute groups |
This behavior happens for OCTETSTRING types where the value is binary data and not strings. Binary data is not translated for display. It is forwarded as the binary data. |
No data in the Tivoli Enterprise Portal I have a script data source and I am not getting any data in the Tivoli Enterprise Portal. |
Change the setting for the trace to: ERROR (UNIT:shell ALL) If you see a trace like the
following, your script is not returning the data in the expected format.
|
Cannot remove agent from the Tivoli Enterprise Portal When you try to remove an agent, you see that the agent is still listed in the desktop navigation view, but it is not available. |
This problem occurs if the agent remains
in the managed system list in IBM Tivoli Monitoring. Perform the following steps:
|
Service or process monitoring returns zeros for metrics When you create an agent to monitor Service availability, the column values are not correct. All metrics, including the Process ID, are zero. For example:
|
Ensure that the agent is being run under an Administrator ID. |
Trying to monitor a non-existent Performance Monitor object The agent runs and tries to monitor a Performance Monitor object, but a status of INACTIVE is displayed along with the error code OBJECT_NOT_FOUND. |
The Performance Monitor object does not exist in the system. Specify a Performance Monitor object that exists in the system, or install the monitored application that creates the Performance Monitor object. |
Trying to monitor a non-existent WMI class The agent runs and tries to monitor a WMI class, but a status of ACTIVE is displayed along with the error code NO_INSTANCES_RETURNED. |
The WMI class was not found and does not exist in the system. WMI collection displays the error message in the log each time the agent tries to collect the data. Specify a WMI class that exists in the system, or install the monitored application. |
Agent runs a command return code but does not return data When your agent runs a command
return code script, the end of the log might display the information
such as the line in the following example:
If this type of information is in the log and is the last to be displayed, the command return code script did not end normally. The agent is locked up and cannot return data. |
Investigate the trace log. An entry similar to this example indicates that the command return code has not run to completion. Amend your command return code so that it is completed in a reasonable amount of time (for example, no more than 10 seconds). |
UNKNOWN status is displayed when you try to monitor a service You try to monitor a service but the message Status=UNKNOWN is displayed, even though you already verified that the agent has Administrator privileges. |
The service is not installed on the system. When the view is being built for the Availability table, rows where the status is Unknown must be filtered out to prevent confusion. Filtering rows out is especially needed when an application is composed of a set of optional services. The lack of a service is not an error in this case; it is normal. |
No data is shown for an attribute group You try to monitor a data source but no data is shown from an attribute group that collects data. |
If the systems used for developing and testing the agent are different, the WMI classes and the Performance Monitor objects can be different too. Develop and test the agent on the same version of Windows and the same version of the monitored application that you want to manage. |
Do not see situations in the console I installed my agent, situations, and workspaces. I configured the agent and started it. I see it in the Tivoli Enterprise Portal, but I do not see the situations in the navigator tree or the Situation Event Console. I checked, and the situations are associated with the correct nodes. |
Restart the Tivoli Enterprise Portal. |
Data being sent by the agent does not look like it can be in the Tivoli Enterprise Portal The data looks like the attributes are not being parsed in the correct places. |
Reinstalling Tivoli Monitoring is the safest way to ensure
that you do not introduce incompatibilities between the phases of
your development.
Note: The agent builder includes features to prevent
you from introducing these types of changes into later versions of
your agent. These features ensure that you do not have this type of
problem as you build and deploy updates to an agent.
|
The status for a reinstalled agent indicates that the agent is configured in Managed Tivoli Monitoring Services When you develop an agent, you likely generate, install, test, update the agent and then generate, install and test again. When you follow these actions, the agent status indicates that the agent is configured in Managed Tivoli Monitoring Services after the agent is reinstalled. |
If you modified the agent configuration, then reconfigure the agent, and restart. |
I installed my agent, situations, and workspaces. I configured the agent and started it. I see it in the Tivoli Enterprise Portal, but I do not see the situations in the navigator tree. They are not associated with the correct node. (I right-click on the node and select the situations, but they do not show in the default list.) |
When you create the situation on a live system, it is automatically distributed to the agent by using the host name. This host name is not available in every environment. So you must distribute the situation to a generic managed system list that exists when the agent is installed. Distribute the situations to the CUSTOM_app_name00 Managed System List, and reimport the situations. Then rebuild the Solution Installer image, and reinstall the Tivoli Enterprise Monitoring Server support for the agent. |
Agent fails One reason this behavior can occur is that the ICCRTE_DIR might not be set in the ENV file on Windows systems. |
cygwin might be installed
so that ICCRTE_DIR is set as:
The Kxxinstall.log has the following text that shows
the problem:
You can fix this problem by taking cygwin out of
the path and setting ICCRTE_DIR manually. |
Subnode names Subnodes from different agent instances have the same managed system name. |
The Managed System Name for a subnode consists
of 2 letter agent Product Code:first 24
characters of the Subnode ID:3 letter Subnode Type The first 24 characters of subnode IDs must be unique for all instances of the subnode type in the IBM Tivoli Monitoring installation. The agent automatically adds a prefix PC to prevent the subnodes from colliding with subnodes created by other agents. It automatically appends the subnode type to prevent collisions with other subnode types in the same agent. It uses the first 24 characters of the Subnode ID (which you control) as the final token. No part of the Agent Instance or the Agent Host System is used in constructing the Subnode Managed System Name. By using the same Subnode ID in 2 instances of the Agent, the Managed System Names collide. This name collision happens even for Agent Instances hosted on separate systems. The result is the subnodes do not function. |
Installing 2 agents with the same script name On Windows operating systems, when two different agents are installed on the same Tivoli Enterprise Monitoring Agent, and they both have a script with the same name, the script from the last agent that was installed or deployed overwrites any existing scripts of the same name. Scripts are copied into the instdir\tmaitm6 directory without a warning. |
There is no solution. The files are copied into the TEMA directory so that they are all in a consistent place and easily accessed by the agent and each other. The situation is the same for Windows and UNIX systems. |
Agent Configuration stops When you deploy or configure an agent that contains subnodes and requires a minimum IBM Tivoli Monitoring version of 6.2.1 on a Tivoli Enterprise Monitoring Agent that has a IBM Tivoli Monitoring version earlier than 6.2.1, configuration might stop. |
|
SNMP attribute group not collecting data reliably Data is collected intermittently or not at all. The SNMP version
and credentials are configured correctly The Performance Object Status
Error Code for the attribute group shows NO RESPONSE RECEIVED.
Note: This issue applies to SNMP attribute groups, so the Object
Type in the Performance Object Status table is SNMP.
The agent trace file shows the following message: Timeout occurred. No response from agent. Here is a sample entry: (48A18C71.000A-12:snmpqueryclass.cpp,1714,"internalCollectData") Timeout occurred. No response from agent. |
The IBM Tivoli Monitoring SNMP data provider is multithreaded to enhance performance.
The SNMP data source that is being monitored might not be able to
respond to multiple incoming requests in a timely manner. There are
tuning options that can improve reliability of data collections:
|
An agent monitoring SNMP V2 events, does not receive traps. The MIB for the monitored device indicates the enterprise OID for the trap ends with a ".0". The received SNMP V2 traps contain the snmpTrapOID.0 varbind with a value of enterprise OID.specific type. Example: The received snmpTrapOID is: 1.3.6.1.4.1.1302.3.8.10.2.1.0.6 Enterprise OID defined in the MIB for this trap: 1.3.6.1.4.1.1302.3.8.10.2.1.0 |
The Agent Builder SNMP event receiver follows RFC 2089 so it can handle SNMP V1 and V2 traps. This RFC states that the last token of the snmpTrapOID.0 varbind is the specific trap field. This token and the preceding token if it is a 0 is removed from snmpTrapOID.0 to create the enterprise field. If your MIB includes enterprise OIDs that end with a .0 but receives traps with the last two tokens that are removed from the snmpTrapOID, varbind does not match the enterprise OID in the MIB. To fix, you must make the following modification to your agent:
|
Windows commands not running as expected Take Action command does not run Windows command as expected. Availability Functionality Test does not run Windows command as expected. Script data source does not run Windows command as expected. |
The mechanism that is used to start a process
in Take Action commands, and agent runtime data providers is the native Windows process management API, CreateProcess(). With this command you can start processes
that are implemented in .bat, .cmd, or .exe files. Windows implements several common command
functions as internal commands in the shell and not as programs. These
include commands like echo and dir (common commands that are used to test script execution). Since
these commands are not programs, they cannot be started with createProcess(). To call these commands, create a .bat or .cmd file that contains the commands. It is possible to start the command
processor and provide the built-in command as an argument, as shown
in the following example:
Remember that you are collecting the return code from cmd, not echo in this example. |
CIM data provider stops collecting data from AIX OpenPegasus 2.6.1 CIM Server. Data collection resumes if the AIX CIM Server is stopped and restarted. |
This issue is resolved in IBM Pegasus CIM Server V2.6.1.35 You can display the current version of the IBM Pegasus CIM Server file sets by using lslpp -l sysmgt.pegasus.cimserver.rte Upgrade the IBM Pegasus CIM Server to 2.6.1.35 or later For current details, see the AIX Common Information Model overview |
Collecting metrics through the Windows APIs |
To collect metrics through the Windows APIs, the agent must be hosted on a Windows operating system. Remote registry administration must be enabled on the remote systems. |
CIM data provider stops collecting data from the Solaris WBEM Server. |
Several Solaris patches are required to achieve
a stable Solaris WBEM CIM Server. Minimum versions that are required
are: Solaris 9:Patch Synopsis
Solaris 10:Patch Synopsis
|
CIM data provider intermittently fails to collect data from Solaris 9 WBEM Server after all patches are applied. |
This issue is reported to Sun Microsystems and
is a vendor limitation. If you are running a typical multi-threaded agent against few remote systems, the agent might send all of the requests concurrently to a single WBEM CIM Server. The CIM Server might not handle all requests. In a typical environment, with multiple remote CIM Servers, the requests are spread out across systems and this problem does not occur. Set the following environment variable in the agent env or inifile: CDP_DP_LOCK_CIM_ACCESS=YES When CDP_DP_LOCK_CIM_ACCESS=YES is set, the agent serializes the CIM requests that it sends. The lock covers the lifespan of the request to the other system. When the request is received, the agent unlocks and then processes the result. Note: This flag can have
a negative effect on the scale of the agent. Use this flag only when
it is necessary.
|
CIM data provider stops collecting data from Solaris 10 WBEM CIM_FileSystem class. The Performance Object Status for this attribute group reports GENERAL ERROR. |
This issue is reported to SUN and is a Vendor
Limitation. There is an error in the Solaris WBEM CIMOM. Requests
for the CIM_FileSystem Class log this error in the WBEM log:
The workaround is to collect data from the Solaris_LocalFileSystem class, or to recycle the WBEM CIM Server. |
CIM data
provider cannot collect data remotely from Solaris 10 WBEM server
after you apply Solaris Patches:
|
If you are running Solaris 10 6/06 or earlier, you must modify your WBEM configuration file to allow
remote connections after you install patch 121308-XX. See the following SunSolve document for detailed information:
|
Local configuration JRE warning Locally configuring an agent displays this warning: Kincfgexit Java Runtime Environment was not detected! Extended agent configuration is disabled - if remote configuration for the agent is supported, complete the process using the Tivoli Enterprise Portal. |
The Monitoring Agent for Windows OS was deployed by using tacmd createnode, so no Java was installed on the local system. The agent is configured
enough to allow it to connect to the Tivoli Enterprise Monitoring Server by using the default agent configuration parameters. Start
the agent to allow it to connect to the Tivoli Enterprise Monitoring Server. You can then complete the configuration by using the Tivoli Enterprise Portal. Optionally,
you can install a supported JRE locally to configure the agent by
using the Manage Tivoli Enterprise
Monitoring Services interface. This problem is known
to occur on 6.2.2 releases. Also, use the steps in the following problem: Password is not stored when locally configuring the agent on a Windows system. |
Password is not stored when locally configuring the agent on a Windows system |
When the Java Runtime Environment is installed by Tivoli Monitoring, a special patch is applied to encrypt the
password. When the JRE is manually installed, this patch is not applied.
If you manually installed the JRE, do the following steps so that
passwords are correctly encrypted:
|
Connecting to Microsoft SQL Server by using integrated authentication is failing |
You can connect to a Microsoft SQL Server without a user ID
or password by using integrated authentication from Microsoft. This only works on Microsoft Windows operating systems and requires that you have a JDBC
driver that supports integrated authentication. Integrated authentication
can be used for JDBC connections in the Agent Builder browser or in
the agent run time.
|
Microsoft SQL Server and JDBC driver compatibility errors with Java Runtime Environment (JRE) version 1.6 |
Microsoft requires you to use the JDBC 4.0 driver JAR file sqljdbc4.jar when you use Java Runtime Environment (JRE) version 1.6. This driver must be used with any JRE version 1.6 or newer. If you use a JRE version 1.5, you must use the older JDBC JAR filesqljdbc.jar. |
JDBC connections are failing to find my JDBC driver JDBC connections are failing to find my JDBC driver that was added to the class path. |
Make sure that the JDBC driver is compatible with the JRE that you are using. JDBC 4.0 drivers are compiled with Java 1.6. To use a JDBC 4.0 driver, the JRE that you configure the agent to use must be at least Java 1.6. Agent Builder uses Java 1.5, so you must use a JDBC driver that is compatible with Java 1.5 to use the JDBC browser in Agent Builder. |
Agent support files remain after you uninstall it. After you uninstall an agent, the agent support files remain on the system. |
The support files for the Tivoli Enterprise Monitoring Server and the Tivoli Enterprise Portal are
not removed by uninstalling the agent.
|
Navigator display in the Tivoli Enterprise Portal shows a combination of old and new nodes or shows the wrong data when you click a node When you remove attribute groups, rename them, or reorganize existing groups into or out of navigator groups, the display in the Tivoli Enterprise Portal might show a combination of old and new nodes. |
If you are using Agent Builder 6.2.1.2 or later, resolve this issue by restarting the Tivoli Enterprise Portal Server and then, restart the Tivoli Enterprise Portal. |
Agent installation fails for agents that are built with Agent Builder V6.2.2 or earlier For agents built with Agent Builder V6.2.2 or earlier, installing the agent might prevent other agents from being successfully installed. If the TECLIB directory does not exist in TMAITM6 when the Agent Builder agent is installed, then a file is created with the name TECLIB. When subsequent agents attempt to create files in the TECLIB directory, the agent installation fails. |
Perform the following steps:
|
Installing the agent or application support fails with error message: KQZ0208E When you install the agent or application support and specifying a password with special characters, the login to the monitoring server fails with error message: KQZ0208E. The specified user name or password is incorrect. |
The password cannot include special characters. |
SDA (self describing agents), Problem uploading support. I am developing an agent by using Agent Builder and I am using SDA (self describing agents). It uploaded my support one time, but it does not seem to be doing it anymore. What can I do? |
You can increment the agent version number. or you can use tacmd deleteappinstallrec -t XX -v version (XX is the two-character product code for the agent. version is an 8 digit long number, such as 06230000). This command deletes the tag that tells the system it is uploaded. |
Ping attribute group on Windows slow to respond It is a long time before I start getting data back from my ping attribute group on Windows. |
Name resolution on Windows can take about 5 seconds to timeout if an IP address does not resolve to a host name. If you defined several devices that have this name resolution issue the ping attribute group takes longer than expected to initialize. To prevent this problem, make sure the devices you define can be resolved in DNS or by using entries in your local host file. |
Authorization failures when you use SSH public key authentication |
Check the agent log file. An error message similar to the following one, indicates that the agent process cannot open the private key file. This error might happen on Windows systems since the agent runs as a service that might not be able to access another users private key file. Resolve this problem by making sure the agent process runs as a user that can read the private key file. The following is an example of an error message when the agent process cannot open the private key file: (4C6D417B.0048-1230:userauth.c,631,"file_read_privatekey") -16 - Unable to initialize private key from file |
Running the installIraAgentTEMS.sh results in an error on UNIX Running the installIraAgentTEMS.sh results in the following error on UNIX: Installation failed. Please see the log in /opt/IBM/ITM/logs/product_code_TEMSInstall.log The end of the product_code_TEMSInstall.log contains the following
lines:
|
Extract your agent package into a directory other than /tmp/product_code, where product_code is in lowercase. |
Agent with SSH to Windows does not run a command An agent with SSH to Windows, does not run a command, or, the agent runs the command with SSH to Windows with no effect. |
To run a remote command to a Windows host, you must have a Linux like shell environment installed. Cygwin is an example of a Linux like shell environment. To verify whether a shell
environment exists, SSH or log on to the remote host and enter the
command:
If the command runs, then a shell environment exists. |
On Windows, a mapped drive to a Ping, log file, JMX, or JDBC configuration file cannot be read On Windows, a mapped drive to a Ping, log file, JMX, or JDBC configuration file cannot be read. The agent runs as a service and cannot see the mapped drives. |
On Windows, do not use a mapped drive to store files that are required by the agent. |
Missing or unexpected data for a socket attribute group in the Tivoli Enterprise Portal There is missing or unexpected data for a socket attribute group in the Tivoli Enterprise Portal. |
Check the agent log if there are missing rows of data for a socket attribute group or if the data is not as expected. If there is missing or unexpected data, check the log even if the Performance Object Status for the attribute group displays NO_ERROR, as NO_ERROR is displayed if any valid rows were returned. |
The HTTP data provider does not properly handle URLs that use the https protocol This problem occurs when you use an IBM Java run time on Solaris or HP-UX systems. You know that you have this problem if the Tivoli Enterprise Portal shows http:// before each of your https URLs. For example, http://https://website.ibm.com |
This problem can be solved in one of 2 ways.
One solution is to add the following parameter to the JVM
arguments runtime configuration property:
Another solution is to use a Java run time that was not provided by IBM if you have one available on the system. |
I am testing my agent in Agent Builder and I do not seem to be getting any results. Performance Object Status shows Timed Out (and possibly some other values that indicate there might be a problem) |
Sometimes the trace settings can slow the agent down in the test mode enough that data is not visible in the builder. This problem is most common in data sources like SOAP, JMX, and JDBC that use a Java provider. Lower your Java Trace Level setting by choosing a value higher in the list. Usually agents work with a trace level of DEBUG_MID. Sometimes it is necessary to set the trace level to ERROR (which is the value that is used when the agent is installed). |
The Tivoli Enterprise Monitoring Server does not start and I receive a message that the number of installation packages exceeds the maximum of 512. |
Install an uninitialized QA1CDSCA definition file that causes the Tivoli Enterprise Monitoring Server to extend the space that is used for application definitions. For more information about obtaining and installing the file, see Installed packages exceed the 512 maximum |
I have an agent that is built
with an Agent Builder version before 6.3. Script data providers that
use SSH are failing with messages similar to this:
|
Rebuild the agent with Agent Builder version 6.3 or later. |
On a Windows Server 2012 system, for an agent with a Internet Control Message Protocol (ICMP) data source, an outbound ping works but an inbound ping fails. |
You must enable the server to respond to ping.
Use the following procedure:
|
On a Windows Server 2012 system, for an agent with a remote script data source, the inbound SSH connection fails. |
You must create an inbound rule for SSH connection.
Use the following procedure:
|
How can I find the available parameters to perform silent command line configuration of an Agent Builder agent? |
Note:
|
Problem | Solution |
---|---|
Cognos data model for an agent is loaded into Framework Manager but testing a table causes a table or view not found error |
This problem can have 2 different causes
and solutions:
|
Testing any of the tables in the IBM_TRAM schema in Framework Manager results in an error on DB2 Testing
any of the tables in the IBM_TRAM schema in Framework Manager results
in following error on DB2:
|
Grant select permission on all the Tivoli Reporting and Analytics
Model tables. Change <IBM_TRAM> to the schema used for Tivoli Reporting and Analytics
Model and <Tivoli Data Warehouse user ID> to the Tivoli Data Warehouse user ID.
|
When you run on Windows, testing an Oracle data source in Tivoli Common Reporting fails When you run on Windows, testing an Oracle data
source in Tivoli Common Reporting
fails with an error similar to:
|
The Oracle client libraries are not in the system PATH environment variable. Edit the PATH environment and add the directory that contains the Oracle client DLL files. Restart the Tivoli Common Reporting server. |
The ManagedSystem dimension cannot be used when you create reports for an agent I am not able to use the ManagedSystem dimension when I create reports for my agent. |
Perform the following checks:
|
Warnings when Model Advisor is run against the Cognos data model When the Model Advisor is run against
the Cognos data model, the
following warnings are shown:
(continued on the next page) |
These warnings are expected and do not signify a problem with the Cognos data model. |
Warnings when Model Advisor is run against the Cognos data model (continued)
|
These warnings are expected and do not signify a problem with the Cognos data model. |