Fingerprint-based discovery relies on Predictive Intelligence for generating suggestions for discovery. If the Application Fingerprints dashboard does not display any suggestions, ensure that the Predictive Intelligence is configured correctly.

1. Ensure that Predictive Intelligence is installed and active.
   1. Navigate to System Definitions > Plugins.
   2. Search for Predictive Intelligence.
   3. Verify that the Predictive Intelligence plugin is installed.

      
      
      

      If Predictive Intelligence is not installed, click the Install button.
   4. Click the Predictive Intelligence tile.
   5. On the Predictive Intelligence form, verify that the Status is Active.
      If the status is Inactive, click the Activate/Repair link under Related Links.
2. Ensure that the scheduled job controlling the fingerprint-based discovery, is enabled and configured correctly.
   1. Navigate to System Definitions > Scheduled Jobs.
   2. Select the Applications suggestion - ITOM Autodisco scheduled job from the list.
   3. Ensure that the Active check box is selected.
   4. Check that the discovery frequency is set to one hour as shown in the following example:

      
      
      

3. Check that the clustering solution responsible for grouping processes into suggestions is active.
   1. Navigate to Predictive Intelligence > Clustering Definitions.
   2. Click the application _suggestion definition.
      The ML Solutions tab displays the ml_x_snc_global_global_application_suggestion solution.
   3. Verify that the State value is Solution Complete and Progress is at 100%.

      
      
      

   4. If the State value is not Solution Complete, configure Predictive Intelligence as described in Configuration tips for Predictive Intelligence.
4. Check that the clustering solution created enough process groups, which become suggestions.
   1. Navigate to System Definitions > Tables and locate the Virtual running processes [v_cmdb_running_process] table.
   2. Check that there are more than 1000 records in this table.
   3. If there are less than 1000 processes, discover more processes by running horizontal discovery on additional IP ranges or subnets.
   4. If there are still less than 1000 processes after discovering additional IP ranges or subnets, disable the fingerprint-based discovery.
5. Check that the Application suggestion - ITOM Autodisco transaction is not slow-loading, as described in View and kill active transaction. If this transaction is running for a long time, kill it.

Adjust definitions of process-based suggestions before starting discovery. Typically, you need to fine-tune automatically generated suggestions if the suggestion accuracy is less that 100%. Predictive Intelligence that creates suggestions also provides the accuracy estimate for each suggestion. The lack of accuracy may result in failure to discover required applications or discovering wrong applications in addition to the correctly discovered applications.

1. Navigate to Discovery > Home.
2. Click Application Fingerprints.
   The Application Fingerprints dashboard opens.
3. Navigate to the list of suggestions by one of the following methods:
   • On the Top Suggestions pie chart, view the top suggestions arranged by the number of processes making up these suggestions.
   • To view only suggestions for application servers, click the Suggested Application Servers tile. The filtered list of suggestions for application servers is displayed. This list is also filtered to display only suggestions whose accuracy is 100%.
   • Alternatively, click All suggestions. The unfiltered list of all suggestions is displayed.
4. In the Application Suggestions list, click the relevant suggestion in the Suggested Group Name column.
   The Application Suggestion form opens.

   
   
   

5. To change the name used for the fingerprint-based application, enter the preferred name in the Name field. For example, many applications run java processes and the suggested group name for these groups is java. You may want to assign a name that better describes the application that you want to discover using this suggestion (for example, java_myweblogic).
6. To change the name used for the CI class, enter the preferred CI class name in the Suggested CMDB CI Class field.
7. To fine-tune the classifier rule, change the regular expressions in the Suggested Classifier Rule field.
   For example, add version information to the regular expression for discovering a Tomcat application: .*tomcat.* -Dopss.*version=12.1.3.*
8. Click Update.
9. If you are not sure how to change the classifier rule, fine-tune it after performing the initial discovery as described in Adjust regular expression used for fingerprint-based discovery.

Fine-tune the regular expression in the classifier rule to make discovery results more precise. Predictive Intelligence extracts regular expressions from processes and assigns them to classifier rules as part of creating suggestions. If you are not satisfied with the discovery results or the suggestion accuracy is lower than 100%, modify the regular expression.

1. Navigate to Application > Module.
2. Click Discover Application.
   The fingerprint-based discovery finds the application and creates a CI class, a classifier, and a pattern.
3. On the Application Suggestions list, click the newly created process classifier.
   The Process Classification form opens.

   
   
   

4. In the Parameters value field under Condition, notice the regular expression automatically selected by Predictive Intelligence. In this example, it is ."-DweblogicName=."
5. Click Padlock icon next to Test with.
6. Click the Look up icon to select the server on which you want to test additional regular expressions.
   The list of all discovered servers opens.
7. Select the sever that runs the application that you want to discover using the customized classifier.
   The selected server appears in the Test with field.
8. Right-click the header of the Process Classification form and select Save.
   The system runs a test and displays the results on the Test results tab at the bottom of the page.

   
   
   

9. On the Test results tab, check the processes that were discovered using the regular expressions configured for this classifier.
10. Adjust the regular expression in the Parameters value field under Conditions. For example, make it more generic: "weblogic."
11. Right-click the header of the Process Classification form and select Save.
    The system runs the test again and displays the results on the Test results tab at the bottom of the page.
12. On the Test results tab, verify that the fingerprint discovery found the applications as you expected.

By default, the fingerprint-based discovery is enabled. Disable application discovery based on processes, if Predictive Intelligence does not generate enough processes.

1. Add the system property sys_property process.clustering.appfingerprint.enable as described in Add a system property.
2. To disable, enter false in the Value field.
3. To enable the fingerprint-based discovery again, change the Value field to true.