Skip to content
AnglinAI Docs

ERR_MARKER_SUBMIT_FAILED — Marker Submit Failed

HTTP Status: 502 Retryable: Yes Automatic retry: The system retries submission up to 2 times with 2-second backoff via callWithRetry. This error surfaces only after all retries are exhausted.

What the User Sees

Document processing service is temporarily unavailable. (ERR_MARKER_SUBMIT_FAILED)

What Causes This Error

The Datalab Marker API (https://api.datalab.to) returned a 5xx error or was completely unreachable when the system attempted to submit a PDF for processing. Marker is used as the primary text extraction engine for PDF documents.

This is typically a transient issue caused by Marker infrastructure downtime, network connectivity problems between Cloudflare Workers and the Marker API, or Marker’s processing queue being overwhelmed.

The submission step is the initial POST request that uploads the PDF to Marker. If the submission itself fails (as opposed to the processing timing out), it usually indicates a more fundamental connectivity or availability issue.

Intermittent failures can also be caused by request size limits — very large PDFs may be rejected by Marker’s upload endpoint or by intermediate proxies.

Resolution Steps

For Users

  1. Wait a few minutes and try again. The processing service may be experiencing temporary issues.
  2. If the error persists, try uploading a smaller version of the document.
  3. Try converting with a different quality tier — some tiers may not use the Marker API.

For Administrators

  1. Check the Marker API status by testing directly:
    Terminal window
    curl -X POST https://api.datalab.to/api/v1/marker \
      -H "X-Api-Key: $MARKER_API_KEY" \
      -F "file=@test.pdf"
  2. Verify the MARKER_API_KEY is valid and has not expired.
  3. Check Grafana for Marker API error rate trends. Look for log entries containing “Marker API” or “datalab.to”.
  4. If Marker is down for an extended period, consider temporarily disabling Marker in the conversion pipeline and relying solely on vision-based conversion.
  5. Check if Datalab has published any status updates or maintenance notices.