The Incidents page groups related events and anomalies into incidents. While the Anomalies and Trends pages show individual data points, each incident collects the events that NVIDIA Minerva correlated, so you can investigate one problem instead of many alerts.
Incidents appear after auto-correlation is configured. See Configure Auto-Correlation and Configure Enhanced Auto-Correlation.
Filter Incidents
The Filter row sets which incidents the view shows:
-
Time range: click the time chip and choose a time range.
-
Filters: click + and select a dimension to filter on, for example
eventType.
Unlike Anomalies views, Incidents views do not have anomaly duration or confidence score filters.
Incident Table
Each row is one incident. The newest incidents appear first.
Incidents do not have an open or closed status. Each incident records the events NVIDIA Minerva correlated in one correlation window.
|
Column |
Description |
|---|---|
|
ID |
A unique identifier for the incident. |
|
Timestamp |
When NVIDIA Minerva grouped the events into the incident. |
|
Correlated Event Count |
The number of events and anomalies in the incident. |
|
Groups |
The topology groups of the devices in the incident. |
|
Roles |
The topology roles of the devices in the incident, for example |
|
Devices |
The devices that reported events in the incident, for example |
|
Potential Impact |
Servers connected to the incident's devices, which might be affected by the problem. Click the cell to see the list. |
|
Root Cause |
NVIDIA Minerva's classification of the likely cause, for example Faulty link, Congestion, or PFC storm. Undetermined means no known pattern matched. The column is empty when no root-cause classification ran for the incident. |
To sort, click the ID, Timestamp, or Correlated Event Count header. Sorting applies to the rows already loaded.
To add, remove, or rearrange columns, click the table's settings icon, or drag a column header to move it. To find a loaded incident by its ID, click the search icon and enter part of the ID. The search is case-sensitive.
When new incidents arrive during a Latest time range, New entries found appears. Click the refresh icon to load them.
Incident Details
Select an incident to see its details below the table. The first incident is selected when the page opens.
-
Topology: the part of the topology involved in the incident, with its anomalies and events. Playback starts automatically and adds the events in the order they happened. Use the back, play/pause, and next controls to step through them. The counter shows which event you are on, for example 3 of 12. To see the anomalies and events for an interface or device, click its label.
-
Events: a table of every event in the incident, with its Event Id, Timestamp, Device Name, If Name, Event Type, and Event Category. Event types such as
HostGpuDriverVersionMismatchorLinkDownedCounterappear in the Event Type column. Select an anomaly event to see its details below the table.
To copy a link to the incident, click Share Link and select Copy Incident URL.
Understanding Correlation
NVIDIA Minerva groups events into incidents with one of two correlation methods, set for each correlation namespace. See Configure Auto-Correlation.
Dimensional Correlation
Dimensional correlation groups events based on shared metadata. For example, a rack correlation incident groups all anomalies occurring within the same physical rack. This is effective for identifying localized hardware issues, power failures, or top-of-rack switch misconfigurations.
Topology-Aware Correlation
Topology correlation groups events based on their relationship in the network fabric. For example, if an upstream spine switch experiences an anomaly and the downstream leaf switches show a spike in pause frames, NVIDIA Minerva correlates these as a single incident to show how the issue propagates.
Incidents Views and Notifications
Incidents views use the same controls as Trends and Anomalies views. See Manage Views. Admins and Editors can create and save views. Incidents views can't be added to a correlation namespace.
To send notifications for new incidents, open the view's settings in the View panel and turn on Notifications. See Enable Notifications on a View. Before you turn on notifications, add an eventType filter to the view. Without one, notifications are turned off when you save the view.
Last updated: