On Mar 31, 2005, at 12:58 PM, Anjan Dave wrote:

> I was able to replicate the pgbench database on the SAME machine to
> pgbenchslave database. What I would like to do next is replicate 
> ANOTHER
> new database from the master to a NEW, separate machine, which will act
> as the slave.
>

assuming that the schema is already copied over to node 3....

what you want to do is create node 3 with "store node" then add the 
listen and path entries for node 3 to nodes 1 and 2.

then issue a "subscribe set" slonik command to tell node 3 from where 
to get the set.  that's pretty much it.

then you sit back and watch it replicate the whole shebang in one big 
transaction.

if you can keep the indexes (other than PK indexes) off the node 3 
until the copy is done, you'll likely save a lot of time by deferring 
the index creation until one bulk operation.


Vivek Khera, Ph.D.
+1-301-869-4449 x806