Hi all;

I'm need a solid solution for both slony as the replication mechanism with a 
connection pooling/load balancing option for incoming connections from a JDBC 
app.

Using pgpool II in master_slave mode I can do this:


1) do a slony switchover

2) alert pgpool II that the master is down either by issueing a pcp detatch 
node command or by stopping the slon daemons and shutting down the (former) 
slony master

My concern with the above is that pgpool II will continue to send writes to 
the original slony master after the switchover is completed until it detects 
that it needs to do a failover itself either via a detatch node command or 
via the health check recognizing that the old master is no longer available

Has anyone else solved this with pgpool II ? If so how?

Has anyone solved this via another tools (say pgbouncer and lvm or pgbouncer 
and HAProxy or something entirely different)? 

I'd like to hear what everyone else ids doing in this area and how well it's 
working. What your remaining hurdles are (if any), etc..

Thanks in advance for any feedback.