Parallel processing on UNIX based systems do not convert each image to PDF

[PierreLeBrun22]

I need to convert a large amount of DOCX to PDF in a minimum of time.
The DOCX to be converted is quite simple, it contains only 2 images and some text.

I'm using a parallel stream to reduce the processing time, it works on Windows but for some reason when I run my application in a docker container on a UNIX system, half of the converted files don't have images.

I use the version 2.0.1