{% extends "_layout.html" %} {% set title = "Those earlier results cannot be used" %} {% set status = "422 Unprocessable Content" %} {% set description = "A continues_from entry names results that cannot be used." %} {% set also_on_a_job = true %} {% block main %}

A run that starts part-way through its flow takes the results of a node it does not run from an earlier job, named in create’s continues_from. One of those entries names results that cannot be used: step and index say which node, and reason says why; at create, job_id says which job.

What to do

  1. Read reason. not_found, not_owner, deleted, not_completed, expired or withheld: those results are not available, so run from an earlier step — one whose results you have — or name a job of your own whose results can be used.
  2. For not_completed, name the job that ran the node: a job that only copied a node’s results in did not run it.
  3. For unsupported, this deployment continues no earlier job: send the node’s outputs in the upload instead, or run the whole flow.
  4. An archived job may be continued from: archived is registered and never raised.

The body

{
  "type":     "https://sc.example.com/server-errors/prior-results-unavailable",
  "title":    "Those earlier results cannot be used",
  "status":   422,
  "detail":   "...",
  "instance": "/v1/jobs",
  "step":     "cts",        // REQUIRED on this type
  "index":    "0",          // REQUIRED on this type
  "job_id":   "01J9...",    // REQUIRED at create; absent while the job stages
  "reason":   "expired",    // REQUIRED on this type
  "trace_id": "4bf92f3577b34da6a3ce929d0e0e4736"
}
Member
step, indexREQUIRED At create, the entry’s node. While the job stages, the node whose results were found unusable: the entry’s own, or one upstream of it in the same earlier job.
job_idREQUIRED at create The job the entry named, as the request sent it. Absent while the job stages: the error is the job’s error, which every reader of the job reads, so it names no job, and the job’s continues_from names the earlier job to each reader who may read it.
reasonREQUIRED not_found — no such job, or one you cannot read: one answer for both, so it confirms nothing. not_owner — a job you can read and do not own. deleted — the job is deleted. not_completed — that job did not complete that node, or only copied it in. expired — the artifact holding the node’s outputs, or its manifest, has been deleted. withheld — an administrator withheld one of them. unsupported — this deployment continues no earlier job, so it refuses any entry at create. archived is registered and never raised. An unrecognised value means act on the type alone.
trace_idoptional The request’s W3C trace id. Quote it in a support request.

detail is prose and may be reworded at any time. Branch on type, and read the members above for the specifics — never parse detail.

Where it comes from

POST /v1/jobs Every entry is checked before anything is uploaded.
GET /v1/jobs/{id} As a rejected job’s error, with no job_id: every entry is checked again while the job stages, and its results can go, or be withheld, in between.
{% include "_help.html" %}

What this is not

archive-rejected with reason: "missing_member" is a node the run reads whose results the client sent neither in the archive nor as a job in continues_from.

{% endblock %}