workflow.json

{
  "$id": "https://open-metadata.org/schema/metadataIngestion/workflow.json",
  "$schema": "http://json-schema.org/draft-07/schema#",
  "title": "MetadataWorkflow",
  "description": "OpenMetadata Ingestion Framework definition.",
  "type": "object",
  "definitions": {
    "sourceConfig": {
      "description": "Additional connection configuration.",
      "javaType": "org.openmetadata.schema.metadataIngestion.SourceConfig",
      "type": "object",
      "properties": {
        "config": {
          "oneOf": [
            {
              "$ref": "databaseServiceMetadataPipeline.json"
            },
            {
              "$ref": "databaseServiceQueryUsagePipeline.json"
            },
            {
              "$ref": "databaseServiceQueryLineagePipeline.json"
            },
            {
              "$ref": "dashboardServiceMetadataPipeline.json"
            },
            {
              "$ref": "messagingServiceMetadataPipeline.json"
            },
            {
              "$ref": "databaseServiceProfilerPipeline.json"
            },
            {
              "$ref": "databaseServiceAutoClassificationPipeline.json"
            },
            {
              "$ref": "pipelineServiceMetadataPipeline.json"
            },
            {
              "$ref": "mlmodelServiceMetadataPipeline.json"
            },
            {
              "$ref": "storageServiceMetadataPipeline.json"
            },
            {
              "$ref": "storageServiceAutoClassificationPipeline.json"
            },
            {
              "$ref": "driveServiceMetadataPipeline.json"
            },
            {
              "$ref": "searchServiceMetadataPipeline.json"
            },
            {
              "$ref": "testSuitePipeline.json"
            },
            {
              "$ref": "metadataToElasticSearchPipeline.json"
            },
            {
              "$ref": "dataInsightPipeline.json"
            },
            {
              "$ref": "dbtPipeline.json"
            },
            {
              "$ref": "applicationPipeline.json"
            },
            {
              "$ref": "apiServiceMetadataPipeline.json"
            },
            {
              "$ref": "reverseIngestionPipeline.json"
            },
            {
              "$ref": "mcpServiceMetadataPipeline.json"
            },
            {
              "$ref": "policyAgentPipeline.json"
            }
          ]
        }
      },
      "additionalProperties": false
    },
    "source": {
      "description": "Configuration for Source component in OpenMetadata Ingestion Framework.",
      "type": "object",
      "properties": {
        "type": {
          "description": "Type of the source connector ex: mysql, snowflake, tableau etc..",
          "type": "string"
        },
        "serviceName": {
          "description": "Type of the source connector ex: mysql, snowflake, tableau etc..",
          "type": "string"
        },
        "serviceConnection": {
          "description": "Connection configuration for the source. ex: mysql , tableau connection.",
          "$ref": "../entity/services/connections/serviceConnection.json#/definitions/serviceConnection"
        },
        "sourceConfig": {
          "$ref": "#/definitions/sourceConfig"
        }
      },
      "additionalProperties": false,
      "required": ["type", "sourceConfig"]
    },
    "processor": {
      "description": "Configuration for Processor Component in the OpenMetadata Ingestion Framework.",
      "type": "object",
      "properties": {
        "type": {
          "description": "Type of processor component ex: pii-processor",
          "type": "string"
        },
        "config": {
          "$ref": "../type/basic.json#/definitions/componentConfig"
        }
      },
      "additionalProperties": false,
      "required": ["type"]
    },
    "stage": {
      "description": "Configuration for Stage Component in the OpenMetadata Ingestion Framework.",
      "type": "object",
      "properties": {
        "type": {
          "description": "Type of stage component ex: table-usage",
          "type": "string"
        },
        "config": {
          "$ref": "../type/basic.json#/definitions/componentConfig"
        }
      },
      "additionalProperties": false,
      "required": ["type"]
    },
    "sink": {
      "description": "Configuration for Sink Component in the OpenMetadata Ingestion Framework.",
      "type": "object",
      "properties": {
        "type": {
          "description": "Type of sink component ex: metadata",
          "type": "string"
        },
        "config": {
          "$ref": "../type/basic.json#/definitions/componentConfig"
        }
      },
      "additionalProperties": false,
      "required": ["type"]
    },
    "bulkSink": {
      "description": "Configuration for BulkSink Component in the OpenMetadata Ingestion Framework.",
      "type": "object",
      "properties": {
        "type": {
          "description": "Type of BulkSink component ex: metadata-usage",
          "type": "string"
        },
        "config": {
          "$ref": "../type/basic.json#/definitions/componentConfig"
        }
      },
      "additionalProperties": false,
      "required": ["type"]
    },
    "logLevels": {
      "description": "Supported logging levels",
      "javaType": "org.openmetadata.schema.metadataIngestion.LogLevels",
      "type": "string",
      "enum": ["DEBUG", "INFO", "WARN", "ERROR"],
      "default": "INFO"
    },
    "workflowConfig": {
      "description": "Configuration for the entire Ingestion Workflow.",
      "type": "object",
      "properties": {
        "loggerLevel": {
          "$ref": "#/definitions/logLevels",
          "default": "INFO"
        },
        "raiseOnError": {
          "description": "Control if we want to flag the workflow as failed if we encounter any processing errors.",
          "type": "boolean",
          "default": true
        },
        "successThreshold": {
          "title": "Success Threshold",
          "description": "The percentage of successfully processed records that must be achieved for the pipeline to be considered successful. Otherwise, the pipeline will be marked as failed.",
          "type": "integer",
          "default": 90,
          "minimum": 0,
          "maximum": 100
        },
        "openMetadataServerConfig": {
          "$ref": "../entity/services/connections/metadata/openMetadataConnection.json"
        },
        "config": {
          "$ref": "../type/basic.json#/definitions/componentConfig"
        }
      },
      "additionalProperties": false,
      "required": ["openMetadataServerConfig"]
    }
  },
  "properties": {
    "id": {
      "description": "Unique identifier that identifies this pipeline.",
      "$ref": "../type/basic.json#/definitions/uuid"
    },
    "name": {
      "description": "Name that identifies this pipeline instance uniquely.",
      "$ref": "../type/basic.json#/definitions/entityName"
    },
    "openMetadataWorkflowConfig": {
      "description": "OpenMetadata Ingestion Workflow Config.",
      "type": "object",
      "properties": {
        "source": {
          "$ref": "#/definitions/source"
        },
        "processor": {
          "$ref": "#/definitions/processor"
        },
        "sink": {
          "$ref": "#/definitions/sink"
        },
        "stage": {
          "$ref": "#/definitions/stage"
        },
        "bulkSink": {
          "$ref": "#/definitions/bulkSink"
        },
        "workflowConfig": {
          "$ref": "#/definitions/workflowConfig"
        },
        "ingestionPipelineFQN": {
          "description": "Fully qualified name of ingestion pipeline, used to identify the current ingestion pipeline",
          "type": "string"
        },
        "pipelineRunId": {
          "description": "Unique identifier of pipeline run, used to identify the current pipeline run",
          "$ref": "../type/basic.json#/definitions/uuid"
        },
        "enableStreamableLogs": {
          "description": "Enable streaming logs to a remote log storage via the OpenMetadata Server",
          "type": "boolean",
          "default": false
        },
        "ingestionRunnerName": {
          "description": "Name of the ingestion runner executing this workflow. Set at dispatch time by the execution environment.",
          "type": "string"
        }
      },
      "required": ["source", "workflowConfig"],
      "additionalProperties": false
    }
  },
  "oneOf": [
    {
      "required": ["name", "openMetadataWorkflowConfig", "source", "sink"]
    },
    {
      "required": ["name", "openMetadataWorkflowConfig", "source", "bulkSink"]
    }
  ],
  "additionalProperties": false
}