Documentation Grafana Cloud Knowledge graph Configure Alerts Amends

Custom amend alert mapping

Amend alerts record when configuration, version, or metadata changes occur in your environment. These alerts don’t represent failures; instead, they provide context for understanding how systems evolve.

When to create an amend alert

Use an amend alert for:

• Application version changes
• Deployment scaling events
• Node replacement or lifecycle transitions
• Feature flag toggles
• Configuration drift that is not yet a failure
• Rolling upgrades or rollbacks

Amend alerts help correlate what changed with what broke.

Required labels

Recommended:

Best practices to write failure alerts

Use the following best practices to help you write amend failure alerts.

Detect label/value transitions using changes()

promql Copy

changes(label_replace(myapp_build_info, "version", "$1", "app_version", "(.+)")[10m]) > 0

Use normalized recording rules for version tracking

YAML Copy

record: asserts:version:info
expr: |
  group by (job, service, asserts_env, asserts_site, asserts_source, version) (
    label_replace(build_info, "version", "$1", "app_version", "(.+)")
  )
labels:
  asserts_source: myapp
  asserts_version_type: service

Avoid using for:

Change events represent discrete transitions, so a waiting period isn’t needed.

Example: Version change

YAML Copy

alert: MyAppVersionChanged
expr: |
  changes(label_replace(myapp_build_info, "version", "$1", "version_label", "(.+)")[10m]) > 0
labels:
  asserts_alert_category: amend
  asserts_entity_type: Service
  asserts_severity: info
  asserts_source: myapp
  asserts_version_type: service
annotations:
  summary: 'MyApp version changed'
  description: 'Detected a version update or rollback.'

Example: Scaling event

YAML Copy

alert: DeploymentReplicaCountChanged
expr: |
  kube_deployment_spec_replicas{deployment="checkout"} 
    != kube_deployment_status_replicas{deployment="checkout"}
labels:
  asserts_alert_category: amend
  asserts_entity_type: Deployment
  asserts_severity: info
  asserts_source: kubernetes
  asserts_version_type: deployment_scale
annotations:
  summary: 'Replica count changed'
  description: 'Replica count for the checkout deployment was updated.'

How amend alerts appear in the knowledge graph

Amends:

• Display as blue bars along an entity’s RCA workbench timeline
• Provide critical context when analyzing failures or performance regressions
• Don’t impact entity health scoring
• Are recorded as a historical timeline of configuration evolution

Amends and failures are often reviewed together. Amends explain what changed; failures explain what went wrong.

Next steps

• To learn how to create alerts, refer to Configure alert rules
• To learn how to import a YAML file for alert creation, refer to Import to Grafana-managed rules