Yes.
And a library as the one on the master have to established on the standby server in order to continue WAL archives on
standby.
As far as I remember I experienced the same situation using the "hot_standby" option in the "standby server" section(together with at riggerfile).
There is no solution that makes automatic switching between replication nodes ensuring no dataloss even not a glusterfs just by itself.
A third replicated server has to be added. PostgreSQL - in my opion - ought be able to meet network interruptions as they
seem to quite common making nodes switch several times each day. As one of you kind people said: To be sure against dataloss
use a cluster(at least 2 servers) plus a replication server.
Thanks a lot for helping.
Poul