arazzo: 1.0.1
info:
  title: Apache Kafka Review Consumer Group Lag
  summary: Resolve the cluster, list the consumer groups, and pull the lag summary and assignment for one group.
  description: >-
    Consumer lag is the primary health signal for a Kafka deployment: a group
    whose lag is growing is falling behind its producers. This workflow resolves
    the cluster, lists the consumer groups so the caller can see the whole
    consuming estate, then reads one group in detail for its state, partition
    assignor, coordinator, member list, and lag summary. Reading the group state
    alongside the lag matters, because a group stuck rebalancing shows lag for a
    completely different reason than one that is simply under-provisioned. 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: kafkaRestProxyApi
  url: ../openapi/kafka-rest-proxy.yml
  type: openapi
workflows:
- workflowId: review-consumer-group-lag
  summary: Inspect a consumer group's state, membership, and lag summary.
  description: >-
    Lists the consumer groups on a cluster and reads one group's coordinator,
    members, state, and lag so a lag alert can be attributed to a cause.
  inputs:
    type: object
    required:
    - consumerGroupId
    properties:
      consumerGroupId:
        type: string
        description: The consumer group id to inspect in detail.
  steps:
  - stepId: resolveCluster
    description: >-
      Resolve the cluster id from the REST Proxy so the review does not depend
      on a hardcoded cluster identifier.
    operationId: listClusters
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      clusterId: $response.body#/data/0/cluster_id
  - stepId: listGroups
    description: >-
      List every consumer group on the cluster, giving the reviewer the full
      consuming estate before narrowing to the group under investigation.
    operationId: listConsumerGroups
    parameters:
    - name: cluster_id
      in: path
      value: $steps.resolveCluster.outputs.clusterId
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      consumerGroups: $response.body#/data
  - stepId: readGroup
    description: >-
      Read the target group in detail. The state distinguishes a healthy STABLE
      group from one stuck in a rebalance, while the lag summary and member list
      show whether the group has enough consumers for its partitions.
    operationId: getConsumerGroup
    parameters:
    - name: cluster_id
      in: path
      value: $steps.resolveCluster.outputs.clusterId
    - name: consumer_group_id
      in: path
      value: $inputs.consumerGroupId
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      groupState: $response.body#/state
      partitionAssignor: $response.body#/partition_assignor
      coordinator: $response.body#/coordinator
      consumers: $response.body#/consumers
      lagSummary: $response.body#/lag_summary
  outputs:
    clusterId: $steps.resolveCluster.outputs.clusterId
    consumerGroups: $steps.listGroups.outputs.consumerGroups
    groupState: $steps.readGroup.outputs.groupState
    lagSummary: $steps.readGroup.outputs.lagSummary