On a 9.6 streaming replica, we do table scans for stats and other things. During these scans, the replication is paused (the 'recovering' postgres process has 'waiting' appended to it). We're not using transactions with these scans. Is there anything we can do to prevent the pausing?
