Kyma module health as metric input

[a-thaler]

Description
A Kyma module by design has an operational health. Also the entities managed by the module are having a health usually. When users want to get an overview of it from a monitoring perspective, they could start collecting the mtrics of a module to gather insights. However, often the metrics are providing insides about the operational health and not so much about the business health. The fact that a ServiceBinding of the BTP Operator is in a bad state will not be reflected. So a user would need to set up kube-state-metrics or alike and start collecting the status of all the involved resources, push them to a backend and start creating dashboards.

Instead Kyma should provide out-of-the-box monitoring capabilities for the modules so that users easily can figure out if some resource is in an unhealthy state. That should be possible in a similar way as it was realized for the telemetry module in #728, so having an API to enable metric collection for any Kyma module.

If enabled, the telemetry module could discover all modules via the moduleTemplates, determine the moduleCR status and all module subresource statuses. In Cloud Logging prepared Dashboards will show the module health.

apiVersion: telemetry.kyma-project.io/v1alpha1
kind: MetricPipeline
metadata:
  name: icke
spec:
  input:
    kyma:
      enabled: true

Tasks

• Preparation
  • Enforce a well-defined contract about the module status to rely on
  • Assure that subresources are listed in the moduleTemplates for all modules
• Implementation

Reasons

Attachments

Release Notes