[Pgpool-general] Redundant pgpool-II nodes?

[Avi Blackmore]

Hi,

	Our current server setup is a main server with web, application and  
database (PostgreSQL 8.3) servers and a RAID unit for mass storage.   
I'm looking to set up a redundant server and RAID unit, colocated at a  
nearby location, to make use of load-balancing and do transparent  
failover in case of problems.  Obviously, we need database replication  
to do this.

	PGPool-II looks like a good solution, since it handles DDL  
replication, is synchronous, and can do multimaster replication.   
Also, it load-balances, which is nice.  But, my concern is that having  
one PGPool server process, running on just one machine,  just makes  
that server process (and its machine) a single point of failure.  This  
defeats the purpose, since my entire goal is to enable 100%  
transparent failover, with no downtime in the case of a single server  
(or RAID unit) failing.

	Ideally, what I would like to do is have TWO PGPool processes, one  
per machine, and have them talk to both backends, and synchronize with  
each other so as to avoid conflicts.  But I'm not certain this is  
possible.  Can two instances of PGPool communicate with each other in  
the way I'm describing?  If not, is there another way to use PGPool to  
achieve what I'm trying to do?
--
Avi Blackmore
Head Programmer/System Administrator
Agri ImaGIS Technologies, Inc.