Skip to content

Fundamentals

The top level of an OORS message describes when and by what the observations were made, optionally which contract they relate to, and carries the list of results.

Any string value MAY contain ${VAR_NAME} references resolved at runtime by tooling; see Variables.

Back to TOC

Example

Minimal

Only the required fields: the envelope, source.process, and one result.

apiVersion: v0.8.1
kind: ObservabilityResults
id: 9f1c8a52-2b4e-4a6e-9d3e-7c2a1f0b4d11
observedAt: "2026-01-15T09:30:00Z"
source:
  process: nightly-quality-suite
results:
  - id: no_dup_rows
    type: check
    name: No duplicate rows
    status: pass
{
  "apiVersion": "v0.8.1",
  "kind": "ObservabilityResults",
  "id": "9f1c8a52-2b4e-4a6e-9d3e-7c2a1f0b4d11",
  "observedAt": "2026-01-15T09:30:00Z",
  "source": {
    "process": "nightly-quality-suite"
  },
  "results": [
    {
      "id": "no_dup_rows",
      "type": "check",
      "name": "No duplicate rows",
      "status": "pass"
    }
  ]
}

Full

A full envelope: source with all fields, a run lifecycle object, a contract reference (including the server it is served on), results, the overall status, tags, and customProperties. (See Results for the result fields and Run Lifecycle for run.) The complete files are in examples/.

apiVersion: v0.8.1
kind: ObservabilityResults
id: 3d2b6e90-7a1c-4f5b-8e2d-1a9c4b6f0e23
observedAt: "2026-01-15T09:30:00Z"
source:
  process: order-processor-dbt
  runId: build-4821
  vendor: soda
run:
  state: completed
contract:
  contractId: 53581432-6c55-4ba2-a65f-72344a91553a
  standard: ODCS
  version: 3.1.0
  server: production
results:
  - id: dup_rows_myTable
    type: check
    name: No duplicate rows
    status: fail
  - id: freshness_myDataProduct
    type: metric
    name: Data should not be stale
    status: pass
status: fail
tags: ['finance', 'sox']
customProperties:
  - property: environment
    value: production
{
  "apiVersion": "v0.8.1",
  "kind": "ObservabilityResults",
  "id": "3d2b6e90-7a1c-4f5b-8e2d-1a9c4b6f0e23",
  "observedAt": "2026-01-15T09:30:00Z",
  "source": {
    "process": "order-processor-dbt",
    "runId": "build-4821",
    "vendor": "soda"
  },
  "run": { "state": "completed" },
  "contract": {
    "contractId": "53581432-6c55-4ba2-a65f-72344a91553a",
    "standard": "ODCS",
    "version": "3.1.0",
    "server": "production"
  },
  "results": [
    { "id": "dup_rows_myTable", "type": "check", "name": "No duplicate rows", "status": "fail" },
    { "id": "freshness_myDataProduct", "type": "metric", "name": "Data should not be stale", "status": "pass" }
  ],
  "status": "fail",
  "tags": ["finance", "sox"],
  "customProperties": [
    { "property": "environment", "value": "production" }
  ]
}

Field Descriptions

Key Type UX label Required Description Example
apiVersion string API Version Yes Version of the OORS schema used. v0.8.1
kind string Kind Yes The type of OORS object. For OORS, always ObservabilityResults. ObservabilityResults
id string (UUID) ID Yes A unique identifier (could be a UUID) identifying the observability results. 9f1c8a52-2b4e-4a6e-9d3e-7c2a1f0b4d11
observedAt string (ISO 8601) Observed At Yes The moment in time when the observations were done (ISO 8601 format). 2026-01-15T09:30:00Z
source object Source Yes Object describing the source from where the results come (e.g., job, pipeline).
source.process string Process Yes The name of the process (job, pipeline, etc.) that made the observations. order-processor-dbt
source.runId string Run ID No An optional unique identifier for the specific run/execution of the process. build-4821
source.vendor string Vendor No An optional field indicating the vendor or tool that produced the observations. soda
run object Run No Run lifecycle object. Absent ⇒ the document is terminal. See Run Lifecycle.
contract object Contract No Object describing the contract that provided the specifications for the observations. Highly recommended for traceability.
contract.contractId string (UUID) Contract ID No A unique identifier identifying the specific contract (ODCS/ODPS) the observations are based on. 53581432-6c55-4ba2-a65f-72344a91553a
contract.standard string Standard No The standard followed by the contract. ODCS
contract.version string Version No The version of the contract (not the standard). 3.1.0
contract.server string Server No The identifier of the server, as defined in the referenced contract (e.g., the server name in the ODCS servers section), against which the observations were made. A contract can be served on multiple servers; this field records which one the results apply to. production
results array Results Yes An array of results. Each entry is a result object (check or metric). May be empty when the run made no observations.
status string Status No The overall outcome of the observations in this document, so consumers do not need to aggregate results themselves. See Overall status. fail
tags array Tags No A list of tags for categorizing the message. See Tags. ['finance', 'sox']
customProperties array Custom Properties No A list of key/value pairs for custom properties. See Custom & Other Properties.

Note: When a contract is provided, all observations in results are expected to be based on the specifications defined in that contract. Observations not related to the contract should be sent in separate OORS messages (without the contract object).

Note: An OORS document is terminal by default — without a run object it is the authoritative record of the run that produced it. Long-running producers can opt in to progressive reporting with the optional run object; see Run Lifecycle.

Overall status

The optional top-level status takes the same values as results[].status. It saves consumers from aggregating the results themselves, and lets a quality gate or a dashboard read one field. When present, it SHOULD equal the outcome derived from results, so the two never disagree:

  1. error if any result is error — part of the run could not be evaluated.
  2. Otherwise fail if any result is fail.
  3. Otherwise warn if any result is warn.
  4. Otherwise pass if any result is pass.
  5. Otherwise skip if any result is skip.
  6. Otherwise unknown, including when results is empty.

Consumers that need the overall outcome of a document without status SHOULD derive it with the same rule. On an interim snapshot (run.state: running), status covers only the results reported so far and is not the run's outcome.