Postgres needed solution for small down time
Hi
I'm looking for solution enable me to setup postgresql cluster with very small down time. Situation now:
2 postgres servers + heartbeat2 + database on the emc via fibre channel.
Works's ok, but switching between two host can take about 5 minutes. So I'm looking for solution to make this:
postgres master_host and database on the emc, replicate data to postgres slave_host (database local). If master fail, ha2 switch slave_host to be master.
I need replica be close real time or real time (possible?)
I test SlonyI - works ok, lag between databases is about 5 - 10 seconds. But now I don't know how say to heartbeat2 to setup slave to master.
I try to use pgpool2, and bucardo but I have problem to configure it (any working tutorial?).
Production is on rhel4, postgres 8.2, tests I make on debian lenny and postgresql 8.3. On production upgrade postgres to new version now is not acceptable.
I need only one master and one slave, and quickly switching between hosts (master -> slave).
Thanks for any advice
|