Description
For example, when the disk starts to have problem (limit of open files reached, for example) in the middle of a message store, but the message can be removed from the log. The problem in this case is that the receiver is stuck in an infinite loop where the log tells it does not know this message yet, but the store indicated there is already a message with this id and fail to store it.
Would be safer to force the store to replace whatever is already there.