{% extends "_layout.html" %} {% set title = "Artifact not approved" %} {% set status = "403 Forbidden" %} {% set description = "The per-object gate said no: withheld, an issue, or a node archive holding something that is; or a continuation from a job whose results nobody may have." %} {% block main %}

You hold every resource the artifact derives from and have accepted every agreement behind it, but one thing is still missing: an approval to download it. Or the artifact is held back by a rule about the object itself: it is withheld, it is an issue testcase, or it is a node archive that holds something that is, or a member deleted on its own — or this deployment does not hand its kind over the API. Three cases have nothing to ask for: the input of a rejected job, or of one whose resource set the server has not recorded, on a deployment with controlled resources; an output built over content supplied from a held copy tied to no resource; and the live log of a running node or job on a deployment that approves logs, where the job’s set holds a resource whose results need an approval or content tied to no resource: a live log has no artifact to approve, and the finished log is read as an artifact once the node ends.

As a rejected job’s error, it is a run continuing from an earlier job that holds content tied to no resource: nobody may have that job’s results, its owner included, so the server fetches none of them for the run and there is nothing to ask for.

What to do

  1. Look at the artifact in the listing: a fetchable: false object with can_request_access: true has a path to approval — open its page from POST /v1/auth/browser with {"artifact_id": "<id>"}, and ask there.
  2. With can_request_access: false there is no request to make. A deployment with a web portal may offer the object there.
  3. For a node archive, fetch what it holds that you can have — its manifest, log and reports are listed on their own.
  4. On a rejected job, run the whole flow rather than continue from that job, or ask the operator about the content it was built from.

The body

{
  "type":     "https://sc.example.com/server-errors/artifact-not-approved",
  "title":    "Artifact not approved",
  "status":   403,
  "detail":   "...",
  "instance": "/v1/jobs/01J9...",
  "trace_id": "4bf92f3577b34da6a3ce929d0e0e4736"
}

detail is prose and may be reworded at any time. Branch on type — never parse detail.

Where it comes from

GET /v1/jobs/{id} A rejected job’s error, before staging fetches anything of an earlier job named in continues_from: that job holds content tied to no resource, whose node archives the gate refuses everyone.
GET /v1/jobs/{id}/artifacts/{artifact_id} The object’s own gate.
GET /v1/jobs/{id}/logs A node’s archived log, held back the same way; or a running node’s or job’s live log, on a deployment that approves logs. A finished node’s …/logs stays a 303: the approval is checked for the artifact it names.
{% include "_help.html" %}

What this is not

entitlement-denied is your account not holding a resource the artifact derives from, and names it.

not-ready is an artifact still being described, and is worth asking about again.

not-found is an artifact whose bytes are gone.

{% endblock %}