arazzo: 1.0.1
info:
  title: Elasticsearch Bulk Load and Verify
  summary: Check the cluster can take writes, push an NDJSON bulk payload, verify the resulting document count, and inspect the index.
  description: >-
    The ingestion path for loading data at volume. The workflow confirms cluster
    health, submits a bulk NDJSON payload, and then verifies the load rather than
    trusting it — because the _bulk endpoint answers 200 even when individual
    items inside the payload failed, so the response body's errors flag and a
    follow-up count are the only honest confirmation. It closes by reading the
    index through the cat API for a human-readable view of document count and
    store size. 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: elasticsearchApi
  url: ../openapi/elasticsearch-openapi.yml
  type: openapi
workflows:
- workflowId: bulk-load-and-verify
  summary: Submit a bulk NDJSON payload and confirm the documents actually landed.
  description: >-
    Verifies cluster health, submits the bulk payload, checks the per-item errors
    flag, counts the documents now matching in the index, and reads the cat
    indices view for the resulting size on disk.
  inputs:
    type: object
    required:
    - index
    - ndjson
    properties:
      index:
        type: string
        description: The index the bulk payload targets, used for verification.
      ndjson:
        type: string
        description: >-
          The newline-delimited bulk body, alternating action and source lines,
          terminated by a trailing newline.
  steps:
  - stepId: checkClusterReady
    description: >-
      Read cluster health before a volume write, so a red cluster stops the load
      rather than absorbing a partial ingest.
    operationId: clusterHealth
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      status: $response.body#/status
      activeShards: $response.body#/active_shards
  - stepId: submitBulk
    description: >-
      Submit the NDJSON payload. A 200 here only means the batch was processed,
      not that every item succeeded, so the errors flag is captured as an output
      for the caller to assert on.
    operationId: bulk
    requestBody:
      contentType: application/x-ndjson
      payload: $inputs.ndjson
    successCriteria:
    - condition: $statusCode == 200
    - context: $response.body
      condition: $.errors == false
      type: jsonpath
    outputs:
      errors: $response.body#/errors
      took: $response.body#/took
      items: $response.body#/items
  - stepId: verifyCount
    description: >-
      Count the documents now present in the index with a match_all query, giving
      an independent confirmation that the bulk load landed.
    operationId: searchDsl
    parameters:
    - name: index
      in: path
      value: $inputs.index
    requestBody:
      contentType: application/json
      payload:
        query:
          match_all: {}
        size: 0
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      documentCount: $response.body#/hits/total/value
  - stepId: inspectIndex
    description: >-
      Read the cat indices view with verbose headers for a human-readable summary
      of the index health, document count, and store size after the load.
    operationId: catIndices
    parameters:
    - name: v
      in: query
      value: true
    successCriteria:
    - condition: $statusCode == 200
    outputs:
      indicesTable: $response.body
  outputs:
    bulkErrors: $steps.submitBulk.outputs.errors
    documentCount: $steps.verifyCount.outputs.documentCount
    indicesTable: $steps.inspectIndex.outputs.indicesTable