I'm running some bulk imports using an SDK script. I'm using DocumentImporter to OCR the documents. However, after it's done processing, the LFOmniOCR process never terminates. The resources they use is negligible, but after you accumulate enough negligible the machine grinds to a halt.

Is there anything I can do to prevent this? Right now I'm running them in batches, and I just pskill all of them after every batch completes.

If it makes a difference, I'm running this part of the process in a highly parallelized manner.