Disclaimer: Found by a coding agent. I have checked the relevant code myself but did not reproduce the overflows in practice.

There are several instances where the bulk/batch writes use too narrow integer types for statistics (namely int instead of eg. (u)int64_t).

For large collections, this may overflow, rendering the statistics unreliable.

Pointers:
write_batch_response_processor.cpp
bulk_write_exec.h
bulk_write_reply_info.h
bulk_write.idl