[Pgpool-general] comments on the following setup

[Christopher Gorge Marges]

Since the replication diff engine is not quite reliable and always
makes false alarms but we could not tolerate any inconsistency between
database, we would like to use another program for replication.  Our
first option is to use pgcluster just for the replication part.  The
result is the following:

server1     server2
    |
  pgpool
   |      \
db1        db2
   |       /
pgreplicate

It seems to work just fine.  We are still currently testing this
setup.  Pgreplicate does not need to be on both servers since pgpool
will switch over to the other db and replication will no longer be
needed.  However, pgpool will be the single point of failure.  If
server1 is shut down, then server2 will also stop.

server1     server2
    |             |
  pgpool    pgpool
   |        X     |
     db1     db2
   |           /
pgreplicate

Here both pgpool will access db1 and db2.  Second pgpool is just for
standby.  Would there be problems on this setup?