Skip to content
Docs

Incident propagation

On this page
  1. How it works
  2. On the map
  3. Example

When a service fails, the services that depend on it are affected too — even if their own sensors are perfectly healthy. Propagation makes this visible on the Service Map, so that you start your investigation at the root cause instead of at the first symptom.

Dependencies and propagation An arrow from A to B means that A depends on B. When Notifications fails, Mobile Backend, which depends on it, is shown as impacted even though its own sensors are healthy. StorefrontHealthyMobile BackendImpactedHealthy itself, but dependson a failing servicePublic APIHealthyNotificationsDownRoot cause:its own sensors failMessage BrokerHealthyA → B : A depends on B
An arrow from A to B means that A depends on B. When Notifications fails, Mobile Backend, which depends on it, is shown as impacted even though its own sensors are healthy.

How it works#

Onagre follows the dependencies you declared:

  • A root cause is a service whose own sensors are failing (Down or Degraded).
  • Every service that depends on it — directly or through other services — is impacted.
  • An impacted service whose own sensors are healthy is shown as Degraded; one that is already failing on its own keeps its own status.

Nothing to configure: the computation is made each time the map is displayed, from the current state of your sensors. Loops between services are handled: each service is counted once.

Propagation is a reading aid. It does not open incidents on impacted services and does not change who is notified: incidents and notifications only come from sensors.

On the map#

Turn on Propagation in the map's toolbar:

Indicator Meaning
Thick red border A root cause that impacts other services.
Dashed orange border, "Impacted by …" An impacted service, with the name of the root cause.
Orange arrow The path through which the failure spreads.

Click a service for more: an impacted service shows Impacted, its Impact source and its active incidents; a root cause shows its Downstream impact — the number of services it reaches.

Service map with propagation on: Notifications has a red border as the failing service, Mobile Backend a dashed orange border as an impacted service, and the edge between them is highlighted.
With Propagation on, the failing service and the services it impacts stand out.

Example#

On Corvane's map, the SMS provider degrades and the Notifications — SMS gateway sensor fails:

  • Notifications is the root cause: one of its sensors is down;
  • Mobile Backend depends on Notifications: it is impacted, although all its own sensors are up;
  • Storefront, which does not depend on Notifications, stays healthy.

The team looks at the SMS gateway, not at the mobile backend.