Is it correct to assume that shutdown mode smart, i.e.
pg_ctlcluster 9.6 main stop
won't work for PostgreSQL standby servers and that I will always have to use shutdown mode fast (or immediate) instead:
pg_ctlcluster 9.6 main stop -m fast
I am assuming so because of this paragrah in pg_ctl(1):
Three different shutdown methods can be selected with the -m option.
“Smart” mode waits
for all active clients to disconnect and any online backup to finish. If the server is in
hot standby, recovery and streaming replication will be terminated once all clients have
disconnected. “Fast” mode (the default) does not wait for clients to disconnect and will
terminate an online backup in progress. All active transactions are rolled back and clients
are forcibly disconnected, then the server is shut down. “Immediate” mode will abort all
server processes immediately, without a clean shutdown. This will lead to a crash-recovery
run on the next restart.
I am hesitating a bit, because I have been used so far in using fast (or immediate) only in cases of relative emergencies.
Best Answer
A WAL receiver is neither an active client, nor an online backup. So a so-called "smart" shutdown executed on the master will not wait indefinitely for the WAL receiver to go away on its own. Rather, it will actively disconnect the WAL receiver. This is pretty much the last thing it does--it waits for the shutdown checkpoint to finish, and streams the WAL containing that checkpoint record over to the replica, and then disconnects it.
If you are concerned about "If the server is in hot standby, recovery and streaming replication will be terminated once all clients have disconnected", that is describing what happens when smart shutdown is initiated against the replica, not against the master. The master is never "in hot standby", that is a status that only a replica can be in.