Configure Feedback

Warning

Feedback is a prototype feature and is not recommended or supported for use in production environments.

Feedback is a supervised machine learning algorithm that creates Situations based on user feedback.

You can enable and configure Feedback in the $MOOGSOFT_HOME/config/moolets/feedback.conf configuration file. After you enable Feedback, you can select and edit the Feedback brain via the System Settings in the Moogsoft AIOps UI.

Refer to Feedback Reference to see all available properties.

Before You Begin

Before you set up Feedback, ensure you have met the following requirements:

  • Your LAMs or integrations are running and Moogsoft AIOps is receiving events.

  • You have configured the Moolet that is the source of the alerts for Feedback. You select the source using the process_output_of property.

Configure Feedback

Edit the configuration file at $MOOGSOFT_HOME/config/moolets/feedback.conf.

See Feedback Reference for a full description of all properties. Some properties in the file are commented out by default.

  1. Provide a name for algorithm:

    • name:Name of the Feedback Moolet.

  2. Configure Feedback's behavior for when it starts and stops running:

    • run_on_startup: Determines whether Feedback runs when Moogsoft AIOps starts.

    • process_output_of: Defines the Moolet sources of the alerts that Feedback processes.

  3. Configure the Feedback algorithm and how it clusters alerts:

    • membership_limit:Maximum number of Situations that an alert can be a member of.

    • scale_by_severity: Feedback treats alerts with a high severity like alerts with a high entropy value.

    • entropy_threshold: Minimum entropy value an alert must have in order for Feedback to include it in a Situation.

    • single_matching: Match alerts to the most suitable neural network.

  4. Configure the neural networks:

    • inputs:Alert attributes that you are interested in matching and want Feedback to consider when learning a Situation.

    • learn_queues: Collection of actions you want to trigger learning.

    • unlearn_queues: Collection of actions that can trigger the removal of a neural network.

    • rating_threshold:Measures when to trigger learning or unlearning.

    • match_strategy: Determines how Feedback matches alerts.

    • precision: Determines how precisely you want to train the brain.

    • tolerance: Determines the degree of error an alert can have to create a new Situation.

    • window: Determines the length of time in seconds that Feedback analyzes alerts and develops a Situation each time it runs.

Restart the Moogfarmd service to activate any changes you make to the configuration file. See Control Moogsoft AIOps Processes for further details.

Example

The following example demonstrates a simple Feedback Sigaliser:

{
        # Moolet
        name                : "Feedback",
        classname           : "com.moogsoft.farmd.moolet.feedback.CFeedback",
        run_on_startup      : false,
        process_output_of   : "MaintenanceWindowManager",
        membership_limit    : 1,
        scale_by_severity   : false,
        entropy_threshold   : 0.0,
        single_matching     : false,
        inputs              : [ "source","description" ],
        learn_queues        : [ "manual_create","rated" ],
        unlearn_queues      : [ "rated" ],
        rating_threshold    : 3,
        precision           : 92.0,          
        tolerance           : 0.01,           
        exact_match         : false,
        window              : 60
}