Hmm, not really unfortunately. What would have been useful is a copy of the pdf that makes ocr run at 100%. Do you happen to have a copy of it? You should be able to find it by searching for the id
It was not a PDF but all 3 resources that were âblockedâ were image/jpg ..
Unfortunately as i have deleted the resources, i donât have the filename anymore .. when we make a DELETE of a resource via API does it keep data somewhere ?
I think i remember the beginning of the file i had checked while i was investigating and it seems to be the one iâm attaching right now.
But i have created a note with it and the OCR worked perfectly on it.(if needed i can send it to you by mail if you need the perfect file as it was on my desktop)
I think what happens is that it processes the resources in a specific order, always the same, and it got stuck on one of the resources, which means it also couldn't process the following ones. If you go in your profile folder, in the "resources" folder, is there a file with this name f2d7...?
yes I have this file : it was a pdf i had added to a note to test if OCR was still working and Joplin instantly managed to extract data from it. This one worked, but after this one, it got blocked on something else. Based on the sources i can see, I donât think there is a âprocess successfulâ log after each process. So I presume based on this screenshot from yesterday :
3 resources were still pending. I have added this pdf, it started the ocr engine and said
Processing resource 1/4 with this good pdf. It worked but before even saying âProcessing resource 2/4â it was blocked and causing 100% cpu. Maybe itâs not even OCR service but the way it tries to get the next resource ?
Sorry to have deleted resources before trying to debug moreâŠ
And just to check, I have tried to add this pdf to a new note right now and it worked again :
Hello @Laurent, i think i have found a way to reproduce my issue.
I discovered this morning that it was the same behaviour as before : 100%cpu on Mac desktop. There were 11 resources to process and nothing moved. I have listed and deleted all of them with API. Nothing seemed wrong with them (png files). But I had a clue : all these png had been added through copy on my mobile. As soon as all of them were deleted, it said that the Process was finished.
I have found the way to reproduce it :
I have iOS version and Mac Desktop version.
On the iOS i have created a note and i have copied a screenshot to it. (not a file but i have made a screenshot then Copy and Delete)
Then i have synchronized with iOS then with Desktop. As soon as the note was synchronized to the desktop it started the OCR process and now CPU is 100% and nothing changes.
Yes I can try to do it. The issue seems so easy to reproduce : just add an image on mobile and synchronize and the desktop OCR becomes blocked .. There is probably a strange thing on my config that causes that issue (E2EE / Joplin Cloud)
