> Pgpool is a real cool project, I really hope it keeps evolving. I have a couple basic/fundamental questions that I hoping to get some clarification on. > > Question 1: If pgpool is used in conjuntion with slony or a another replication it can be used only for its loadbalancing and failover. I am correct like this? What else needs to be done to set in failover? > > set the backends.... > > backend_host_name =master > secondary_backend_host_name=slave > > set it to lb on and replication off... > > load_balance_mode=true > replication_mode=false > > set the health checks for failover. > > health_check_timeout=20 > health_check_period=20 > health_check_user=user You need to set master_slave_mode to true. > Question 2: In the case that the master goes down pgpool would failover to the slave/secondary. What would be the logical steps to bring it back to normal. Can this be down with no downtime with both? > 1) Turn off health check or weight the weight_master =0 so when you bring the master back load does not go to it when you bring the db backup. or does pgpool leave it down automatically until you restart pgpool? Yes. > 2) Shutdown replication on the master. Bring back the master backup. > > 3) rsynch the data from the slave to the master??? > > 4) turn replication? It seems these steps OK. But you are going to use Slony-I with pgpool, no? Then you might want to sync DBs using Slony-I? > Question 3: Is there anyway to make pgpool loadbalance certain select queries to the slave such as queries to a particular table? Is that a future feature planned? Since pgpool does not parse queries, probably it would not happen in the future. Ideas are welcome how to know a query is accessing particular table without parsing it. -- Tatsuo Ishii ---------------------------(end of broadcast)--------------------------- TIP 8: explain analyze is your friend
