OCR on Papermerge does not seem to finish

[kamal-meeheelaul]

Hello Everyone,
I am facing an issue with Papermerge v3.2. I have uploaded multiple documents using the CLI. All documents are uploaded successfully, however, the OCR part seems to be running for quite a while now. Its on the 5th day, and the CPU usage is constantly 100%. Previously I have loaded around 3,000 files, and the COR was completed within 2 days. This time I loaded around 800 documents of almost small to medium size.

I checked the table ocr_document, and only 2 documents are in STARTED phase, once these documents are completed, other documents are just added in the STARTED phase. Why is this so? Does Papermerge re-run OCR on completed documents again? Any tips on speeding up the OCR? Or if I can disable it temporarily and enable it only at night.

Please help, I have set up a live system and my users are currently unable to use the platform due to the server unavailability (access is too slow due to the high CPU usage by the OCR process).

Thanks.