Anomaly Feedback
Last updated
Last updated
The Augtera anomaly feedback feature enables operators to provide feedback on the operational relevancy of a particular anomaly and label the criteria for that feedback. This enables Augtera to recommend the operational relevance of new anomalies that match the criteria of prior feedback. Augtera populates the recommendation in a new dimension called recommendedCategory. This recommendation can then be used for additional workflows or to determine which anomalies generate notifications to ticketing systems.
Augtera anomalies can be consumed via notifications, dashboards, and views specific to anomalies. The first step in the process of providing anomaly feedback is to select a particular anomaly you want to provide feedback on. Feedback is anomaly object specific. For example feedback given for an anomaly object related to traffic utilization would not necessarily be applicable for an anomaly object related to CPU utilization. The dimension that labels the anomaly object is type. Each defined anomaly object has a different value in the type dimension.
In the example workflow below, we are going to give feedback on a specific anomaly detected on a specific device and interface. The anomaly is associated with the anomaly object called interfaceIn that detects traffic utilization anomalies. We are using a view in the Anomalies page of the UI that is filtered to display only that anomaly object (type=interfaceIn) in the filter.
If the above columns are missing from the table in your view you can add them by selecting the gear icon on the table, then selecting Feedback and Recommended Category columns and then ok. This will add these columns to the table.
Select the feedback Icon for the anomaly you want to provide feedback on.
The anomaly feedback wizard will appear on the screen. See descriptions below for each section in the wizard
This is the anomaly object name. It will match one of the type(s) selected in the view
This allows you to toggle seeing prior anomaly feedback for this anomaly object.
This is the anomaly ID. Note that you can provide feedback on a specific anomaly only once. You can however provide feedback on different anomalies associated with the same anomaly object. Feedback can be modified later if needed by selecting that same anomaly and then the feedback icon. Every anomaly in Augtera is unique and will have a unique ID.
This is the content section and will populate with appropriate controls as the user goes through the wizard. In this example, the user will select the relevance of the anomaly.
This shows the past feedback for this anomaly object in a ranked order. Users can adjust the rank or delete from this screen. When a new anomaly is generated by Augtera it will match the highest ranked past anomaly feedback provided by you on anomalies that are associated with the same anomaly object type . At the bottom of the feedback ranking, which is a different shade, is the current feedback.
Select the appropriate feedback category that categorizes the operational relevance of this anomaly. In this workflow example, we are selecting Operationally relevant & urgent. Select Next to Continue.
Select the dimensions that influenced your feedback. Only dimensions with non null values will appear on this list. A search is available if you know the dimension you want to use. You can select more than one dimension, but a minimum of one is required. In this example we will select the role and MetricDown dimensions. Future anomalies that you want to leverage this feedback will need to match the same dimensions selected in this step.
Select the metrics that influenced your feedback. In the example workflow, there is only one metric available since only a single metric had an anomaly (ifHcInOctets). The value of that metric during the anomaly time period will be displayed. If the anomaly object has multiple measures you may have more options to choose. Once a metric is selected you will need to input the lower and upper bound values. These values also must be matched in a future anomaly to generate a recommended category. In this example we are setting the low end of 0 and the high end to 300 Kb/s. If a new anomaly that is of the type interfaceIn is generated and the value of ifHcInOctets is greater than 0 but but below 300Kb/s and the category dimension is local-major then Augtera will apply a recommended category of Operationally Relevant to it.
Augtera populates the dimension recommendedCategory when a new anomaly matches a prior anomaly feedback. Operators can use this dimension in the Anomaly and Trends views to filter anomaly results and drive specific behavior.
A practical example is to have all anomalies that match feedback of operationally relevant and urgent to generate a service now ticket or slack notification to a specific team.
In this example, we have a view that uses a filter to limit results to only operationally relevant anomalies. This view is configured to send any matching anomalies to a specific slack channel.
