Prometheus Expire a Silence Early

arazzo: 1.0.1
info:
  title: Prometheus Expire a Silence Early
  summary: Find an active silence, inspect what it mutes, expire it, and confirm it is gone.
  description: >-
    The other half of the incident lifecycle: the fix has shipped, and the
    silence needs to come off before its scheduled end so the alert can page
    again if the problem returns. The workflow finds the silence by matcher,
    inspects it to confirm the right one is about to be lifted, expires it, and
    reads it back to prove the state moved to expired. Every step spells out its
    request inline so the flow can be read and executed without opening the
    underlying OpenAPI description.
  version: 1.0.0
sourceDescriptions:
- name: alertmanagerApi
  url: ../openapi/prometheus-alertmanager-api-openapi.yml
  type: openapi
workflows:
- workflowId: expire-silence
  summary: Expire an active silence and verify the alert can fire again.
  description: >-
    Expiring a silence sets its end time to now; it does not delete it. The
    silence stays visible in listings as an audit record of what was muted, by
    whom, and why.
  inputs:
    type: object
    required:
    - silenceMatcher
    properties:
      silenceMatcher:
        type: string
        description: >-
          Label matcher identifying the silence to lift (e.g.
          alertname="HighRequestLatency").
  steps:
  - stepId: findSilence
    description: >-
      List the silences whose matchers cover this label matcher and take the
      first. Silences in the expired state are still returned here, which is why
      the next step checks the state before expiring anything.
    operationId: listSilences
    parameters:
    - name: filter
      in: query
      value:
      - $inputs.silenceMatcher
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      silences: $response.body
      silenceId: $response.body#/0/id
    onSuccess:
    - name: silenceFound
      type: goto
      stepId: inspectSilence
      criteria:
      - context: $response.body
        condition: $.length > 0
        type: jsonpath
    - name: noSilenceFound
      type: end
      criteria:
      - context: $response.body
        condition: $.length == 0
        type: jsonpath
  - stepId: inspectSilence
    description: >-
      Read the silence in full before lifting it. The comment and createdBy
      fields are what confirm this is the silence from this incident and not a
      long-standing maintenance window someone else owns.
    operationId: getSilence
    parameters:
    - name: silenceID
      in: path
      value: $steps.findSilence.outputs.silenceId
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      silenceId: $response.body#/id
      state: $response.body#/status/state
      matchers: $response.body#/matchers
      createdBy: $response.body#/createdBy
      comment: $response.body#/comment
      endsAt: $response.body#/endsAt
  - stepId: expireSilence
    description: >-
      Expire the silence by setting its end time to now. Alerts matching it
      resume notifying on the next Alertmanager evaluation.
    operationId: deleteSilence
    parameters:
    - name: silenceID
      in: path
      value: $steps.inspectSilence.outputs.silenceId
    successCriteria:
    - condition: $statusCode == 200
  - stepId: verifyExpired
    description: >-
      Read the silence back and confirm its state is now expired rather than
      active or pending.
    operationId: getSilence
    parameters:
    - name: silenceID
      in: path
      value: $steps.inspectSilence.outputs.silenceId
    successCriteria:
    - condition: $statusCode == 200
    - condition: $response.body#/status/state == "expired"
    outputs:
      state: $response.body#/status/state
      endsAt: $response.body#/endsAt
  outputs:
    silenceId: $steps.inspectSilence.outputs.silenceId
    previousState: $steps.inspectSilence.outputs.state
    finalState: $steps.verifyExpired.outputs.state
    expiredAt: $steps.verifyExpired.outputs.endsAt