Click + to expand to see the known issues in each release.

Known issues in V1.3.6

For a list of known issues in version 1.3.6, see the Release Notes in the IBM Operations Analytics Predictive Insights Knowledge Center.

Known issues in V1.3.5

For a list of known issues in version 1.3.5, see the Release Notes in the IBM Operations Analytics Predictive Insights Knowledge Center.

Known issues in V1.3.3

Using the documented procedure to change the database user password causes console integration to stop working

Symptom:
After the password update, the snowflake icon is no longer visible in Dashboard Application Services Hub.

Cause
The password is not updated in the blazeConfig.xml file.

Resolution:

Complete the following steps to copy the new password from the taspConfig.xml file, which is updated with the new password, to the blazeConfig.xml file:

1. Edit the <PI_UI_HOME>/wlp/usr/servers/piserver/taspConfig.xml file.

where <PI_UI_HOME> is typically /opt/IBM/scanalytics/UI.

2. Locate the line beginning with: properties.db2.jcc databaseName.

3. Go to the password field and copy the encrypted password that is enclosed in inverted commas.

4. Edit the <PI_UI_HOME>/wlp/usr/servers/piserver/blazeConfig.xml file.

where <PI_UI_HOME> is typically /opt/IBM/scanalytics/UI.

5. Locate the line beginning with: properties.db2.jcc databaseName field.

6. Go to the password field and paste the encrypted password, that you copied from the taspConfig.xml file, within the inverted commas.

7. Save the blazeConfig.xml file.

8. Enter the following command to restart the User Interface:

<PI_UI_HOME>/bin/pi.sh -restart

where <PI_UI_HOME> is typically /opt/IBM/scanalytics/UI.

Cannot mediate data that has epoch time stamps when the data type of the time stamp is set to Number in the Mediation Tool

Symptom:
When you deploy a model, the KPI count shows a value of zero and no data is extracted when you run extraction.

Cause
The data type for the metric group’s timestamp is set to Number in the Mediation Tool.

Resolution:
1. In the Mediation tool, open the Model Properties tab.

2. In the Model Objects pane, expand Model and expand the metric group.

3. Under the metric group, select Timestamp.

4. In the Model Object Properties pane, click the Number value in the Data Type field.

5. In the drop-down list box, select Integer.

6. Repeat steps 2 to 5 for each metric group.

7. Save and redeploy the model.

Note. When you redeploy the model, you receive the following message: Timestamp format, currently integer, is invalid for Timestamp Component. You can ignore this message and choose OK to proceed with the model deployment.

Extraction errors occur when prepared statements disabled for MYSQL database source

Symptom:
When you run the run_extractor_instance command, you see errors similar to the following:

com.ibm.tivoli.netcool.pa.mediation.MediationException: ERR_DB_DATABASE_ERROR
Caused by: com.ibm.tivoli.netcool.pa.mediation.MediationException: ERR_DB_DATABASE_ERROR
at com.ibm.tivoli.netcool.pa.mediation.source.db.DatabaseReader.executeSelect(DatabaseReader.java:390)
at com.ibm.tivoli.netcool.pa.mediation.source.db.DatabaseReader.read(DatabaseReader.java:159)
Caused by: com.ibm.tivoli.netcool.pa.mediation.MediationException: ERR_DB_DATABASE_ERROR
at com.ibm.tivoli.netcool.pa.mediation.source.db.DatabaseReader.executeSelect(DatabaseReader.java:390)
at com.ibm.tivoli.netcool.pa.mediation.source.db.DatabaseReader.read(DatabaseReader.java:159)
Caused by: java.lang.RuntimeException: Use of Statement not supported

Cause:
Prepared statements are disabled for the data source.

Resolution:
1. In the Mediation Tool, open the model that contains the MYSSQL database source.

2. Click the Connection Details tab.

3. Click the Use Prepared Statement check box.

4. From the menu bar, choose File->Save.

5. To redeploy the model, from the menu bar, choose Predictive Insights->Deploy Model.

Upgrading to Operations Analytics- Predictive Insights 1.3.3 can lead to more alarms being generated

Symptom:
After you upgrade to version 1.3.3 from an earlier version, Operations Analytics – Predictive Insights generates more alarms than expected.

Cause:
The upgrade procedure doesn’t set the correct value for the robustBounds.threshold property. After you execute the following command to check the value of the robustBounds.threshold property, an incorrect value of 0.997 is returned:

$PI_HOME/bin/admin.sh show -t=<topic> -h | grep robustBounds.threshold

Resolution:
Complete the following steps to update the value of the robustBounds.threshold property to 3 and restart the Analytics for each topic:

As the analytics user, typically scadmin, repeat the following steps for each topic:-
1. Change to the $PI_HOME/bin directory.
2. Enter the following command to update the value of the robustBounds.threshold property:

./admin.sh set -t=<topic> robustBounds.threshold 3 -f

3. Enter the following commands to stop and restart Analytics for a topic:

./stop.sh -t=<topic>
./start.sh -t=<topic>

4. Repeat steps 2 and 3 for each topic.

5. Enter the following command to start extraction:

./admin.sh run_extractor_instance -l=<latency>

Installation of Operations Analytics Predictive Insights fails

Symptom:
Installation of Operations Analytics Predictive Insights fails

Cause:
There was insufficient disk space for the Installation Manager to complete the installation.

Resolution:
Increase the amount of free disk space on the server.

Installation Manager crashes

Symptom:
The Installation Manager Crashes when you run it after it has been updated by the OMNIbus core installer.

Cause:
This crash is caused by a known issue with Cairo graphics library issue.

Resolution:
Add the following line to the Installation Manager ini (IBMIM.INI) file:

-Dorg.eclipse.swt.internal.gtk.cairoGraphics=false

The prerequisite scanner shows incorrect results

Symptom:
The prerequisite scanner returns a fail result incorrectly for:

  • Any library package that has a later version than the expected version
  • The PortsInUse:50000 property if port 50000 is in use. The scanner expects port 50000 to be unused.
  • The numLogicalCPU property when it finds the expected number of processors

Resolution:
You can ignore these incorrectly reported failures.

Consolidated alarm shows a different number of children than mentioned in the summary when opened using ViewChildAlarms

Symptom:
The number of child alarms shown when using the ViewChildAlarms function is different to the number specified in the summary.

Cause:
This can occur when the system.alarm.autoclear property is set to ‘false’.

Resolution:
Set the system.alarm.autoclear property to ‘true’.


Consolidated events cannot be displayed in the User Interface if the node name contains an apostrophe

Symptom:
A consolidate event cannot be displayed in the User Interface.

Cause:
The node name in the consolidated event contains an apostrophe.

Resolution:
Open child alarms either individually or collectively.

Resources cannot be searched for in Anomaly Search if the resource name contains an apostrophe

Symptom:
A Resource name cannot be searched for in Anomaly Search if the resource name contains an apostrophe.

Cause:
The resource name contains an apostrophe.

Resolution:
Search on the time stamp instead of the resource name.

Extractor instance crashes due to insufficient memory

Symptom:
The Extractor instance crashes due to insufficient memory.

Cause:
Failure to set the DATABASE_MEMORY parameter for the DB2 database may leave insufficient memory for the Analytics component.

Resolution:
On a server that has both DB2 and the Analytics component installed, ensure that the the DB2 DATABASE_MEMORY parameter is set and that this memory is included when calculating the overall memory requirements for the server. URL to tuning Wiki.

Error validating Jazz installation

Symptom:
The following error is displayed when validating the Jazz installation:

ls: cannot access /home/scadmin/IBM/JazzSM/profile/bin/wsadmin.sh: No such file or directory
Directory /home/scadmin/IBM/JazzSM does not appear to be Jazz Home

Cause:
The Jazz Profile was created in a separate folder to the Jazz installation folder.

Resolution:
Review the folder locations supplied in the DASH parameters section of the UI install process, supplying the correct paths for both JazzSM and the JazzSM profile folder.

Executing run_extractor_instance doesn’t extract data

Symptom:
When you execute run_extractor_instance, no data is extracted.

Cause:
The location of a file system data source is set with a relative path in the Mediation Tool.

Resolution:
Replace the relative path to the file system data source with an absolute path.

ERR_EXTRACTION_PREVIEW_FAILED error when attempting to preview data for a metric group

Symptom:
ERR_EXTRACTION_PREVIEW_FAILED error when attempting to preview data for a metric group that is defined against a MSSQL database table.

Resolution:
When specifying the connection details for a database data source, enter the database name in both the Database and Schema fields in the Connection Details dialog.

Error occurs in the Mediation Tool when performing an extraction preview for a metric group

Symptom:
You see the following errors in the Mediation Tool:

Results: ERR_EXEC_READER
!ENTRY com.ibm.tivoli.pa.model.ui.nature 4 0 2013-04-03 09:54:58.649
!MESSAGE Unexpected Exception
!STACK 0
java.util.concurrent.ExecutionException: com.ibm.tivoli.netcool.pa.mediation.model.validate.ModelValidationException: ERR_EXEC_READER
at java.util.concurrent.FutureTask$Sync.innerGet(FutureTask.java:233)
at java.util.concurrent.FutureTask.get(FutureTask.java:94)

Cause:
Your computer may have gone into hibernate mode while the Predictive Insights Mediation Tool was open. When you start your computer and try to do a metric group content preview this error occurs.

Resolution:
Restart the Mediation Tool.

Separate data sources cannot contain metric groups with the same name

Symptom:
When you deploy a new data source model, you get a message stating that the “Model validation failed.” The details of the error message state that metrics of the same name are defined for more than one data source.

Cause:
The Operations Analytics – Predictive Insights validation requires that all metric groups are unique to a data source.

Resolution:
If you have two or more data sources that contain metric groups with the same name, then you will need to rename the metric groups so that they are unique for each data source.

Timeouts in the Mediation log

Symptom:
Mediation processing has stopped and there is a ERR_EXTR_TIMEOUT seen in the log file, PI_HOME/log/<topic>/Analytics<topic>_log_DataSourceOperator.log

Cause:
This can occur when processing backlog data in an undersized system, or a system that has a set of metric groups with large variation in sizes.

Resolution:
If you are still processing backlog mode data, re-start the data processing. To re-start the data processing from the point where it stopped, run:

$PI_HOME/bin/admin.sh run_extractor_instance -l=<latency>

You only need to specify the -l parameter if mediation will switch into steady-state mode. Optionally, you can specify an end time.

If you are extracting data in steady-state mode when the ERR_EXTR_TIMEOUT message occurs, contact IBM support.

NOTE: if the ERR_EXTR_TIMEOUT message is seen repeatedly after re-starting the data processing, then look for the latest interval for any metric group in the $PI_HOME/var/spool/topics/<topic>/extracted folder. In this case, you may need to start the data processing beginning with the next interval. This can drop up to 10 intervals of data for some metric groups. You will see the data that was dropped in the …/extracted/bad folder after your run:
$PI_HOME/bin/admin.sh run_extractor_instance -s=<start time> -l=<latency>

Optionally, you can specify an end time.

You only need to specify the -l parameter if mediation will switch into steady-state mode. The start time, as stated, should be for the next interval that does not have any data in the extracted folder.

Timeouts in the Mediation log

Symptom:
When processing data in backlog mode, Mediation processing stops and there is a timeout error in the Mediation log, PI_HOME/log/<topic>/Analytics<topic>_log_DataSourceOperator.log

Resolution:
If Mediation is suspended for training and you only see one instance of the ERR_EXTR_TIMEOUT_STOP error and there are no subsequent ERR_EXTR_TIMEOUT_CANCEL errors, which indicate that Mediation has permanently stopped, you can ignore this error. When training completes, mediation automatically resumes. If mediation does not resume, see the resolution for addressing timeouts in the Mediation log in the previous known issue above.

Probe process continually displays as ‘defunct’

Symptom:
The probe process continually displays as defunct when you execute:
ps -ef | grep nco_p_stdin

Resolution:
Run the probe manually to check the output returned.
cd $PI_HOME/probe/omnibus/probes/

./nco_p_stdin
Perform the two checks below if you see an error similar to the following:

/opt/IBM/scanalytics/analytics/probe/omnibus/probes/linux2x86/nco_p_stdin: error while loading shared libraries: libstdc++.so.5: wrong ELF class: ELFCLASS64
1. Ensure that /usr/lib64:/usr/lib is added to the LD_LIBRARY_PATH variable.
echo $LD_LIBRARY_PATH
2. Run the following command to verify that the prerequisite packages are installed.
ls /usr/lib*/libstdc++.so.[56] This command should list four files:
/usr/lib64/libstdc++.so.5

/usr/lib64/libstdc++.so.6

/usr/lib/libstdc++.so.5

/usr/lib/libstdc++.so.6
If all of these files are not listed, ensure that you install all prerequisites.
Note: If you see the following warning when you run the probe manually, it should run successfully when executed by the system. However, you may first need to manually kill the defunct processes.
Warning: W-ETC-004-066: Failed to load properties file /home/scadmin/scanalytics/analytics/probe/omnibus/probes/linux2x86/stdin.props: No such file or directory.

Setting the OMNIbus NCHOME variable can result in errors

Symptom:
Setting NCHOME may cause the probe to point to the wrong location and not run.

No Predictive Insights alarms appear in the AEL.

You may see that the nco_p_stdin process on the Analytics component server has status of “defunct”.
Cause:
This environment variable specifies the home location for Tivoli Netcool/OMNIbus.
It is not mandatory that you manually set this environment variable before running the Tivoli Netcool/OMNIbus installation program.
Resolution:
If your OMNIbus and your Predictive Insights Analytics component are on the same server AND owned by the same user, do not set this variable. This is in line with OMNIbus documentation – search for “Setting Tivoli Netcool/OMNIbus environment variables (UNIX and Linux)”.

Error message when connecting to datasource
Symptom:
The following information messages may be seen in the Mediation tool when you add a datasource:

INFO Cannot connect to datasource [,com.ibm.tivoli.netcool.pa.dao.ConnectionException: ERR_DB_COULD_NOT_CONNECT]

INFO Unable to connect to a datasource, Loading default configuration com.ibm.tivoli.netcool.pa.config.ConfigException: ERR_CONFIG_INST_NOT_FOUND

Cause:
These messages are displayed by default with a first time connection to any datasource.

Resolution:
These messages can be ignored.

Warning that can safely be ignored in Installation logs

Symptom:
The following warning can be seen when you view the installation logs in Installation Manager.
“Warning : : No “conClass” attribute in “panel” element of panel”

Resolution:
This kind of warning can be safely ignored.

Symptom:
The following warning can be seen:

20150607 17:48:10 : echo password XXXXXX
20150607 17:48:10 :
/tmp/runCmdInsatller1433717290215_44: line 4: unexpected EOF while looking for matching `”
/tmp/runCmdInsatller1433717290215_44: line 5: syntax error: unexpected end of file

Resolution:
This kind of warning can be safely ignored

Known issues in V1.3.2

Installation of Operations Analytics Predictive Insights fails

Symptom:
Installation of Operations Analytics Predictive Insights fails

Cause:
There was insufficient disk space for the Installation Manager to complete the installation.

Resolution:
Increase the amount of free disk space on the server.

Installation Manager crashes

Symptom:
The Installation Manager Crashes when you run it after it has been updated by the OMNIbus core installer.

Cause:
This crash is caused by a known issue with Cairo graphics library issue.

Resolution:
Add the following line to the Installation Manager ini (IBMIM.INI) file:

-Dorg.eclipse.swt.internal.gtk.cairoGraphics=false

Consolidated alarm shows a different number of children than mentioned in the summary when opened using ViewChildAlarms

Symptom:
The number of child alarms shown when using the ViewChildAlarms function is different to the number specified in the summary.

Cause:
This can occur when the system.alarm.autoclear property is set to ‘false’.

Resolution:
Set the system.alarm.autoclear property to ‘true’.


Consolidated events cannot be displayed in the User Interface if the node name contains an apostrophe

Symptom:
A consolidate event cannot be displayed in the User Interface.

Cause:
The node name in the consolidated event contains an apostrophe.

Resolution:
Open child alarms either individually or collectively.

Resources cannot be searched for in Anomaly Search if the resource name contains an apostrophe

Symptom:
A Resource name cannot be searched for in Anomaly Search if the resource name contains an apostrophe.

Cause:
The resource name contains an apostrophe.

Resolution:
Search on the time stamp instead of the resource name.

Extractor instance crashes due to insufficient memory

Symptom:
The Extractor instance crashes due to insufficient memory.

Cause:
Failure to set the DATABASE_MEMORY parameter for the DB2 database may leave insufficient memory for the Analytics component.

Resolution:
On a server that has both DB2 and the Analytics component installed, ensure that the the DB2 DATABASE_MEMORY parameter is set and that this memory is included when calculating the overall memory requirements for the server. URL to tuning Wiki.

Error validating Jazz installation

Symptom:
The following error is displayed when validating the Jazz installation:

ls: cannot access /home/scadmin/IBM/JazzSM/profile/bin/wsadmin.sh: No such file or directory
Directory /home/scadmin/IBM/JazzSM does not appear to be Jazz Home

Cause:
The Jazz Profile was created in a separate folder to the Jazz installation folder.

Resolution:
Review the folder locations supplied in the DASH parameters section of the UI install process, supplying the correct paths for both JazzSM and the JazzSM profile folder.

Executing run_extractor_instance doesn’t extract data

Symptom:
When you execute run_extractor_instance, no data is extracted.

Cause:
The location of a file system data source is set with a relative path in the Mediation Tool.

Resolution:
Replace the relative path to the file system data source with an absolute path.

ERR_EXTRACTION_PREVIEW_FAILED error when attempting to preview data for a metric group

Symptom:
ERR_EXTRACTION_PREVIEW_FAILED error when attempting to preview data for a metric group that is defined against a MSSQL database table.

Resolution:
When specifying the connection details for a database data source, enter the database name in both the Database and Schema fields in the Connection Details dialog.

Error occurs in the Mediation Tool when performing an extraction preview for a metric group

Symptom:
You see the following errors in the Mediation Tool:

Results: ERR_EXEC_READER
!ENTRY com.ibm.tivoli.pa.model.ui.nature 4 0 2013-04-03 09:54:58.649
!MESSAGE Unexpected Exception
!STACK 0
java.util.concurrent.ExecutionException: com.ibm.tivoli.netcool.pa.mediation.model.validate.ModelValidationException: ERR_EXEC_READER
at java.util.concurrent.FutureTask$Sync.innerGet(FutureTask.java:233)
at java.util.concurrent.FutureTask.get(FutureTask.java:94)

Cause:
Your computer may have gone into hibernate mode while the Predictive Insights Mediation Tool was open. When you start your computer and try to do a metric group content preview this error occurs.

Resolution:
Restart the Mediation Tool.

Separate data sources cannot contain metric groups with the same name

Symptom:
When you deploy a new data source model, you get a message stating that the “Model validation failed.” The details of the error message state that metrics of the same name are defined for more than one data source.

Cause:
The Operations Analytics – Predictive Insights 1.3.2 validation requires that all metric groups are unique to a data source.

Resolution:
If you have two or more data sources that contain metric groups with the same name, then you will need to rename the metric groups so that they are unique for each data source.

Timeouts in the Mediation log

Symptom:
Mediation processing has stopped and there is a ERR_EXTR_TIMEOUT seen in the log file, PI_HOME/log/<topic>/Analytics<topic>_log_DataSourceOperator.log

Cause:
This can occur when processing backlog data in an undersized system, or a system that has a set of metric groups with large variation in sizes.

Resolution:
If you are still processing backlog mode data, re-start the data processing. To re-start the data processing from the point where it stopped, run:

$PI_HOME/bin/admin.sh run_extractor_instance -l=<latency>

You only need to specify the -l parameter if your mediation will switch into steady-state mode. Optionally, you can specify an end time.

If you are extracting data in steady-state mode when the ERR_EXTR_TIMEOUT message occurs, contact IBM support.

NOTE: if the ERR_EXTR_TIMEOUT message is seen repeatedly after re-starting the data processing, then look for the latest interval for any metric group in the $PI_HOME/var/spool/topics/<topic>/extracted folder. In this case, you may need to start the data processing beginning with the next interval. This can drop up to 10 intervals of data for some metric groups – you will see the data that was dropped. in the …/extracted/bad folder. after your run:
$PI_HOME/bin/admin.sh run_extractor_instance -s=<start time> -l=<latency>

Optionally, you can specify an end time.

You only need to specify the -l parameter is only needed if your mediation will switch into steady-state mode. The start time, as stated, should be for the next interval that does not have any data in the extracted folder.

Timeouts in the Mediation log

Symptom:
When processing data in backlog mode, Mediation processing stops and there is a timeout error in the Mediation log, PI_HOME/log/<topic>/Analytics<topic>_log_DataSourceOperator.log

Cause:
This can occur because model training did not complete within the configured timeout interval, which is set to 10 times the value of the system.aggregation.interval property. This can occur because PI fails to stop counting the timeout period when mediation is suspended due to training being run.

Resolution:
Run the following command to increase the timeout interval:

./admin.sh set -t=<topic> system.topic.timeout.intervals <no. of intervals> -f

set <no. of intervals> to a number of aggregation intervals that is greater than the time it takes for training to complete.

Probe process continually displays as ‘defunct’

Symptom:
The probe process continually displays as defunct when you execute:
ps -ef | grep nco_p_stdin

Resolution:
Run the probe manually to check the output returned.
cd $PI_HOME/probe/omnibus/probes/

./nco_p_stdin
Perform the two checks below if you see an error similar to the following:

/opt/IBM/scanalytics/analytics/probe/omnibus/probes/linux2x86/nco_p_stdin: error while loading shared libraries: libstdc++.so.5: wrong ELF class: ELFCLASS64
1. Ensure that /usr/lib64:/usr/lib is added to the LD_LIBRARY_PATH variable.
echo $LD_LIBRARY_PATH
2. Run the following command to verify that the prerequisite packages are installed.
ls /usr/lib*/libstdc++.so.[56] This command should list four files:
/usr/lib64/libstdc++.so.5

/usr/lib64/libstdc++.so.6

/usr/lib/libstdc++.so.5

/usr/lib/libstdc++.so.6
If all of these files are not listed, ensure that you install all prerequisites.
Note: If you see the following warning when you run the probe manually, it should run successfully when executed by the system. However, you may first need to manually kill the defunct processes.
Warning: W-ETC-004-066: Failed to load properties file /home/scadmin/scanalytics/analytics/probe/omnibus/probes/linux2x86/stdin.props: No such file or directory.

Setting the OMNIbus NCHOME variable can result in errors

Symptom:
Setting NCHOME may cause the probe to point to the wrong location and not run.

No Predictive Insights alarms appear in the AEL.

You may see that the nco_p_stdin process on the Analytics component server has status of “defunct”.
Cause:
This environment variable specifies the home location for Tivoli Netcool/OMNIbus.
It is not mandatory that you manually set this environment variable before running the Tivoli Netcool/OMNIbus installation program.
Resolution:
If your OMNIbus and your Predictive Insights Analytics component are on the same server AND owned by the same user, do not set this variable. This is in line with OMNIbus documentation – search for “Setting Tivoli Netcool/OMNIbus environment variables (UNIX and Linux)”.

Error message when connecting to datasource
Symptom:
The following information messages may be seen in the Mediation tool when you add a datasource:

INFO Cannot connect to datasource [,com.ibm.tivoli.netcool.pa.dao.ConnectionException: ERR_DB_COULD_NOT_CONNECT]

INFO Unable to connect to a datasource, Loading default configuration com.ibm.tivoli.netcool.pa.config.ConfigException: ERR_CONFIG_INST_NOT_FOUND

Cause:
These messages are displayed by default with a first time connection to any datasource.

Resolution:
These messages can be ignored.

Warning that can safely be ignored in Installation logs

Symptom:
The following warning can be seen when you view the installation logs in Installation Manager.
“Warning : : No “conClass” attribute in “panel” element of panel”

Resolution:
This kind of warning can be safely ignored.

Symptom:
The following warning can be seen:

20150607 17:48:10 : echo password XXXXXX
20150607 17:48:10 :
/tmp/runCmdInsatller1433717290215_44: line 4: unexpected EOF while looking for matching `”
/tmp/runCmdInsatller1433717290215_44: line 5: syntax error: unexpected end of file

Resolution:
This kind of warning can be safely ignored

Known issues in V1.3.1

Installation fails when you use the same username to install the Predictive Insights UI and JazzSM/DM on different servers

Symptom:
The Predictive Insights User Interface installation fails when you install the Predictive Insights UI on a different server than JazzSM/Dash, using the same username on both servers.

Resolution:
Use a different username to install the Predictive Insights UI and JazzSM/Dash if you are installing the applications on separate servers. If you have to use the same user, contact IBM.

Summary information not displayed in an alarm for an unpegged metric

Symptom:
In the Active Event list, the summary field is empty for alarms generated for an unpegged metric.

Cause:
There is no expected value for alarms for unpegged metrics and so the summary field is left blank.

Resolution:
In the Model Properties tab, select the metric you wish to unpeg. In the Model Object Properties pane, click the Metric Type field and select Accumulation from the drop-down menu.

Mediation Tool crashes when you open model with a text editor

Symptom:
In the Mediation Tool, when you right click a model and choose Open with a Text Editor, the model configuration screen displays briefly and the Mediation Tool then crashes.

Cause:
This crash is caused by a Cairo graphics library issue.

Resolution:
Add the following line to eclipse.ini:

-Dorg.eclipse.swt.internal.gtk.cairoGraphics=false

Installation Manager crashes

Symptom:
The Installation Manager Crashes when you run it after it has been updated by the OMNIbus core installer.

Cause:
This crash is caused by a known issue with Cairo graphics library issue.

Resolution:
Add the following line to the Installation Manager ini (IBMIM.INI) file:

-Dorg.eclipse.swt.internal.gtk.cairoGraphics=false

Consolidated alarm shows a different number of children than mentioned in the summary when opened using ViewChildAlarms

Symptom:
The number of child alarms shown when using the ViewChildAlarms function is different to the number specified in the summary.

Cause:
This can occur when the system.alarm.autoclear property is set to ‘false’.

Resolution:
Set the system.alarm.autoclear property to ‘true’.

Consolidated events cannot be displayed in the User Interface if the node name contains an apostrophe

Symptom:
A consolidate event cannot be displayed in the User Interface.

Cause:
The node name in the consolidated event contains an apostrophe.

Resolution:
Open child alarms either individually or collectively.

Resources cannot be searched for in Anomaly Search if the resource name contains an apostrophe

Symptom:
A Resource name cannot be searched for in Anomaly Search if the resource name contains an apostrophe.

Cause:
The resource name contains an apostrophe.

Resolution:
Search on the timestamp instead of the resource name.

Validation of single sign on domain fails during Predictive Insights User Interface install

Symptom:
When installing the Predictive Insights User Interface, the validation of the domain you specify in the Single Sign On field fails

Cause:
The fully qualified domain name is specified after the host name in /etc/hosts. For example:
192.168.1.100 server1 server1.ibm.com

Resolution:
Insert the fully qualified domain name before the hostname in /etc/hosts. For example:
192.168.1.100 server1.ibm.com server1

Extractor instance crashes due to insufficient memory

Symptom:
The Extractor instance crashes due to insufficient memory.

Cause:
Failure to set the DATABASE_MEMORY parameter for the DB2 database may leave insufficient memory for the Analytics component.

Resolution:
On a server that has both DB2 and the Analytics component installed, ensure that the the DB2 DATABASE_MEMORY parameter is set and that this memory is included when calculating the overall memory requirements for the server. URL to tuning Wiki.

Error validating Jazz installation

Symptom:
The following error is displayed when validating the Jazz installation:

ls: cannot access /home/scadmin/IBM/JazzSM/profile/bin/wsadmin.sh: No such file or directory
Directory /home/scadmin/IBM/JazzSM does not appear to be Jazz Home

Cause:
The Jazz Profile was created in a separate folder to the Jazz installation folder.

Resolution:
Review the folder locations supplied in the DASH parameters section of the UI install process, supplying the correct paths for both JazzSM and the JazzSM profile folder.

Executing run_extractor_instance doesn’t extract data

Symptom:
When you execute run_extractor_instance, no data is extracted.

Cause:
The location of a file system data source is set with a relative path in the Mediation Tool.

Resolution:
Replace the relative path to the file system data source with an absolute path.
ERR_EXTRACTION_PREVIEW_FAILED error when attempting to preview data for a metric group
Symptom:
ERR_EXTRACTION_PREVIEW_FAILED error when attempting to preview data for a metric group that is defined against a MSSQL database table.

Resolution:
When specifying the connection details for a database data source, enter the database name in both the Database and Schema fields in the Connection Details dialog.

Error occurs in the Mediation Tool when performing an extraction preview for a metric group

Symptom:
The following error can occur within the Mediation Tool:
Results: ERR_EXEC_READER

!ENTRY com.ibm.tivoli.pa.model.ui.nature 4 0 2013-04-03 09:54:58.649

!MESSAGE Unexpected Exception

!STACK 0

java.util.concurrent.ExecutionException: com.ibm.tivoli.netcool.pa.mediation.model.validate.ModelValidationException: ERR_EXEC_READER

at java.util.concurrent.FutureTask$Sync.innerGet(FutureTask.java:233)

at java.util.concurrent.FutureTask.get(FutureTask.java:94)

Cause:
Your computer may have gone into hibernate mode while the Predictive Insights Mediation Tool was open. When you start your computer and try to do a metric group content preview this error occurs.

Resolution:
Restart the Mediation Tool.

Separate data sources cannot contain metric groups with the same name

Symptom:
When you deploy a new data source model, you get a message stating that the “Model validation failed.” The details of the error message state that metrics of the same name are defined for more than one data source.

Cause:
The Operations Analytics – Predictive Insights 1.3 validation requires that all metric groups are unique to a data source.

Resolution:
If you have two or more datasources that contain metric groups with the same name, then you will need to rename the metric groups so that they are unique for each datasource.

Timeouts in the Mediation log

Symptom:
Mediation processing has stopped, and there is a ERR_EXTR_TIMEOUT seen in the log here:
PI_HOME/log//Analytics_log_DataSourceOperator.log

Cause:
This can typically happen when processing backlog data in an undersized system, or a system that has a set of metric groups with large variation in sizes.

Resolution:
If you are still processing backlog mode data, re-start the data processing, and reach steady-state.

To re-start the data processing from the point where it stopped, just run:
$PI_HOME/bin/admin.sh run_extractor_instance -l=, optionally with an endtime.
The -l parameter is only needed if your mediation will switch into steady-state mode.

If you are in steady-state mode when the ERR_EXTR_TIMEOUT message is see, contact IBM support.

NOTE: if the ERR_EXTR_TIMEOUT message is seen repeatedly after re-starting the data processing, then look for the latest interval for any metric group in the $PI_HOME/var/spool/topics//extracted folder. In this case, it may be required to start the data processing beginning with the next interval. This can drop up to 10 intervals of data for some metric groups – you will see that data in the …/extracted/bad folder after running the next command:
$PI_HOME/bin/admin.sh run_extractor_instance -s= -l=
Optionally with an endtime. The -l parameter is only needed if your mediation will switch into steady-state mode. The , as mentioned, should be for the next interval that has not had any data appear in the extracted folder.

Probe process continually displays as ‘defunct’

Symptom:
The probe process continually displays as defunct when you execute:
ps -ef | grep nco_p_stdin

Resolution:
Run the probe manually to check the output returned.
cd $PI_HOME/probe/omnibus/probes/

./nco_p_stdin
Perform the two checks below if you see an error similar to the following:

/opt/IBM/scanalytics/analytics/probe/omnibus/probes/linux2x86/nco_p_stdin: error while loading shared libraries: libstdc++.so.5: wrong ELF class: ELFCLASS64
1. Ensure that /usr/lib64:/usr/lib is added to the LD_LIBRARY_PATH variable.
echo $LD_LIBRARY_PATH
2. Run the following command to verify that the prerequisite packages are installed.
ls /usr/lib*/libstdc++.so.[56] This command should list four files:
/usr/lib64/libstdc++.so.5

/usr/lib64/libstdc++.so.6

/usr/lib/libstdc++.so.5

/usr/lib/libstdc++.so.6
If all of these files are not listed, ensure that you install all prerequisites.
Note: If you see the following warning when you run the probe manually, it should run successfully when executed by the system. However, you may first need to manually kill the defunct processes.
Warning: W-ETC-004-066: Failed to load properties file /home/scadmin/scanalytics/analytics/probe/omnibus/probes/linux2x86/stdin.props: No such file or directory.

Setting the OMNIbus NCHOME variable can result in errors

Symptom:
Setting NCHOME may cause the probe to point to the wrong location and not run.

No Predictive Insights alarms will appear in the AEL.

You may see that the nco_p_stdin process on the Analytics component server has status of “defunct”.
Cause:
This environment variable specifies the home location for Tivoli Netcool/OMNIbus.
It is not mandatory that you manually set this environment variable before running the Tivoli Netcool/OMNIbus installation program.
Resolution:
If your OMNIbus and your Predictive Insights Analytics component are on the same server AND owned by the same user, do not set this variable. This is in line with OMNIbus documentation – search for “Setting Tivoli Netcool/OMNIbus environment variables (UNIX and Linux)”.

Error message when connecting to datasource
Symptom:
The following information messages may be seen in the Mediation tool when you add a datasource:

INFO Cannot connect to datasource [,com.ibm.tivoli.netcool.pa.dao.ConnectionException: ERR_DB_COULD_NOT_CONNECT]

INFO Unable to connect to a datasource, Loading default configuration com.ibm.tivoli.netcool.pa.config.ConfigException: ERR_CONFIG_INST_NOT_FOUND

Cause:
These messages are displayed by default with a first time connection to any datasource.

Resolution:
These messages can be ignored.

Warning that can safely be ignored in Installation logs

Symptom:
The following warning can be seen when you view the installation logs in Installation Manager.
“Warning : : No “conClass” attribute in “panel” element of panel”

Resolution:
This kind of warning can be safely ignored.

Information on the type of relationship discovered for a related metric is duplicated

Symptom:
When a related metric is part of a related events relationship with multiple targets in a consolidated alarm, the details of the information appears more than once

Resolution:
The duplicated information can be ignored.