I'm also seeing conflicts in the pre-archive a lot, especially when I'm pushing DICOMs in bulk. We're using CTP as a buffer between the sending party and XNAT. Data is pushed sequentially to CTP, albeit one after the other (so without waiting between subjects or sessions). Whether CTP splits that up in multiple processes I don't know. I suppose not. But either way, XNAT often chokes on some sessions even so.
It was suggested to me (DevOps people who run the national XNAT/Health-RI infrastructure for the Netherlands) that conflicts might occur when data is coming in via multiple threads and XNAT is waiting for one session to finish, while another interrupts the process. Or something along those lines. Not sure whether that's the case on our side though.
I would be really interested in some more information regarding these errors and how to (a) reliably reproduce them and (b) what to do about them, e.g., explicitly waiting a few seconds after each session?
Any tips are very welcome!
Cheers,
Ralph