Errors and retries
Website API errors use a JSON response with an error message. Validation
responses may include issues, containing field paths and messages.
HTTP status codes
Section titled “HTTP status codes”| Status | Meaning | Action |
|---|---|---|
200 |
Successful read or save | Use the returned revision or result |
201 |
Resource created | Retain its ID and revision |
202 |
Preview/publication job accepted | Poll the job or article publication |
400 |
Invalid content or unsupported fields | Inspect error and field issues |
401 |
Missing, expired, invalid or revoked website credential | Check or replace the credential |
403 |
Missing permission, refused origin or website ownership mismatch | Check the scope and use a server-side request |
404 |
Resource unavailable on this website, or unknown endpoint | Read the current website inventory |
405 |
Unsupported method or operation | Check the endpoint and HTTP method |
409 |
Revision conflict, owned URL, unresolved publication or unavailable production publication | Read current state and reconcile the cause |
413 |
Request too large | Reduce the content or image upload size |
503 |
A required publishing dependency is unavailable | Check the response and retry after the dependency is restored |
The API bounds content requests to 2,000,000 bytes. Images have a separate upload limit; see Images.
Stale revisions
Section titled “Stale revisions”If a save returns 409 with a revision-change message, fetch the current page
or article and compare its content. Apply your intended change against that
revision. Avoid automatically overwriting someone else’s newer content.
Network retries
Section titled “Network retries”Reads can be retried. For page jobs and article previews, retry the identical
request with the same operationId after an uncertain HTTP response. A different
operation UUID requests another operation.
After a save times out, read the current resource and its history to determine whether the revision was committed before submitting another save. After an article creation timeout, check the inventory and retry with its original operation identity.
For a job marked uncertain, keep the job ID and contact PulpSense before
requesting another upload. The publisher retains the deployment attempt and
uses its marker to establish the outcome.