Why am I getting 200 OK "Internal Server Error" when retrieving Form Recognizer Results?

Question

Good evening!

I'm afraid I can't attach the PDFs publicly due to data protection (but could share them in a private message). I am trying to figure out what is causing the sporadic errors we are seeing lately. It only happens once every about 10,000 documents but we cannot work out why.

We receive PDFs from the supplier and almost all work but 1 in 10k fail. I have been checking against the form recognizer limitations but I cannot find out what is going wrong. Below is a comparison of a working PDF and a not working one.

If anybody has any suggestions of what could be wrong, please let me know! No matter what we do or how we trigger the API, it will accept the document on the analyze and the retrieve results always gives internal error. I can't figure out how this PDF differs to the 9,999 others we get the same. We regularly get this error about twice a week where 2 of our 20k docs does this error and we can't work out why.

Thanks!

Emily

Document 1 - works

• .pdf
• 47kb
• 8.27 inches by 11.69 inches
• 1 page
• Uses the same model as the one that doesn't work (on the same training data set)

Document 2 - doesnt work

• .pdf
• 55kb
• 8.5 inches by 11.00 inches
• 4 pages
• Uses the same model as the one that works (on the same training data set)
• Not password locked

example error:

{
    "status": "failed",
    "createdDateTime": "2023-01-16T16:21:37Z",
    "lastUpdatedDateTime": "2023-01-16T16:21:37Z",
    "error": {
        "code": "InternalServerError",
        "message": "An unexpected error occurred.",
        "details": [
            {
                "code": "InternalServerError",
                "message": "An unexpected error occurred."
            }
        ]
    }
}