Metric Health Notifier
Beta

Reading time: 0 minute(s) (0 words)

Currently in beta, the Metric Health Notifier feature allows Nobl9 users to be automatically notified whenever their SLOs stop collecting data for more than 15 minutes (“No data” anomaly notifications). Leveraging these notifications, Nobl9 users can quickly go to SLOs affected by this anomaly and take simple steps to fix problems with their data source or query.

Scope of support

In the beta version, Nobl9 can send the “No data” anomaly notifications via two alert methods:

Opsgenie
Slack
Email

Image 1: Slack notification for the "No data" anomaly.

tip

The remaining alert methods support for this feature is coming soon.

Enabling notifications for the “No Data” anomaly

You can activate the “No data” anomaly notifications on the Nobl9 Web:

For new SLOs: in Step 5 of the SLO wizard:

Image 2: Enabling "No data" anomaly notification in the SLO wizard
For existing SLOs: by editing your SLO in Step 5 of the SLO wizard:

Image 3: Enabling "No data" anomaly notification - Edit SLO step

tip

You can also configure "No data" anomaly notifications for your SLOs, using sloctl. See section below for more details.

Assumptions for anomaly notifications

Anomaly notifications and alerts

“No Data” anomaly notifications and alerts

The underlying reason for receiving a notification about “No data” is not an alert detected by Nobl9, but the detected anomaly.

Consequently, notifications about “No data” anomalies are not the kind: Alert objects in the Nobl9 platform. As such, you can’t retrieve them through the sloctl get alerts command.

Anomaly notifications and SLO objectives

Remember that while you can configure the “No data” anomaly notifications at an SLO level, you will receive a separate notification for each SLO objective that hasn’t collected data for the past 15 minutes (similarly to how alerts work for SLOs).

Maximum number of alert methods per notification

You can add up to five alert methods for a single “No data” anomaly notification.

Anomaly notifications and RBAC

Anomaly notifications and alert methods share the same RBAC rules. This means that you can use data sources from different projects for SLOs with activated anomaly notifications.

To configure the “No data” anomaly notifications for your SLOs, you must have at least one of the following roles in the project that your SLO sits in:

Project Owner
Project Editor
Project Integration User

tip

Suppose you’re a project editor in project-x. Due to your RBAC role, you have access to slo-x which is in project-x. In slo-x, the Notify me on no data is activated, but it’s configured with alert-method-y hooked into project-y, where you don’t have project-level permissions.

If you want to set up an alert-method-y for your notification in slo-x, you must have the Project Owner/Editor/Integration User permissions in project-y.

Anomaly notifications and YAML

SLOs with activated "No data" notifications have an additional object called anomalyConfig in their YAML definition:

- apiVersion: n9/v1alpha
  kind: SLO
  metadata:
    name: my-test-slo
    project: anomaly-testing
  spec:
    alertPolicies: []
    anomalyConfig:
      noData:
        alertMethods:
        - name: notify-on-no-data
          project: dynatrace
        - name: test-anomaly-notifications
          project: default

You can configure the "No data" anomaly notifications through the sloctl apply command, by adding the following definition to your YAML file (see beta limitations below for more details on anomalyConfig):

[...]
    anomalyConfig:
      noData:
        alertMethods:
        - name: notify-on-no-data
          project: dynatrace
        - name: test-anomaly-notifications
          project: default

Beta limitations

Currently, the period for not receiving any data/receiving partial data is hard-coded to 15 minutes.
The anomalyConfig configuration option via YAML is a beta feature that may be subject to change in the future.
Only three alert methods are supported in the beta version. You can add a maximum of five alert methods per “No data” anomaly notification: Opsgenie and Slack.
All clear isn’t supported for “No data” anomaly notifications.