Occasionally, when a resident tries to rez a no-copy item from their inventory during heavy database load or other gridwide issues, this item will fail to rez completely and they will lose their only copy of it. It seems that the rez request removes the item from the resident's inventory (as it should) because it's no-copy, but somehow the item never actually makes it in-world. Maybe this is because the simulator times out while attempting to retrieve the asset data from the asset server. Whatever the cause, this results in content loss with absolutely no recourse for the resident. This could be alleviated by designing some kind of transactional system, in which a failure to retrieve the asset to rez in-world "rolls back" the transaction, returning the item to the owner's inventory. I've seen repeated anecdotes of this issue in the past couple of weeks, so it's clear to me that there's a design flaw here that needs to be investigated. MISC-80 gives an example call for help which would really be better suited for LL Support, but it hilights the underlying issue.