One trick to deal with the first problem is sending logical decoding messages to yourself. That keeps the retained WAL low. Another potentially useful thing is

One trick to deal with the first problem is sending logical decoding messages to yourself. That keeps the retained WAL low. Another potentially useful thing is temporary replication slots which clean themselves up on connection loss. I use those when I don’t need all of the changes. There is also configuration for setting the maximum WAL that’s retained so you don’t murder the server.