You can technically have all of the slons on another server, or all on 
the master, by specifying the conn_info for the node as a remote server.
However slony documentation, as well as myself recommend running a slon 
for each cluster node on the server that hosts the database. The slon 
communicates with the database often.

So, best practice, you need to setup a slon-dbname.conf on each 
postgresql server, that connects to the database on that server, and run 
the slon for the cluster on the master, and then also on each replica.


Alex Haugg wrote:
> hi,
>
> i have a question:
> you starts the slon processes for the master and the slaves on one
> machine (on the master) for all machines?
>
> thank you for the answer.
>
> Am Dienstag, den 18.03.2008, 09:13 -0400 schrieb Nicholas Kiraly:
>   
>> When you use slonik scripts to create your replicated tables, you can 
>> run them through slonik on any machine that can login as the slony users 
>> to all of the database servers. Here's a breakdown of how I usually do it:
>>
>>
>> Setup your slony users on each database server.
>>
>> Setup your slon-dbname.conf on each database server.
>>
>> If your slonik scripts don't, create the matching schema for your 
>> replicated tables on each database.
>>
>> Run your slonik scripts to initialize the slony installation for the 
>> databases on the master. When you do this, slonik will connect to the 
>> replicas as necessary to create the slony schema in those databases.
>>
>> With the slony schema installed, then you can start the slon process for 
>> each database, on each machine. If you use SYNC( ID = source); and then 
>> a WAIT FOR EVENT to confirm in your slonik to confirm config and 
>> replication, you will need to start the slon on each machine before your 
>> script will exit cleanly.
>>
>>
>> If you watch the logs, you'll see a bunch of activity until all the 
>> replicas are up to date.
>>
>>
>>
>> Alex Haugg wrote:
>>     
>>> hi, i have a question:
>>>
>>> in every examples you have the master and the slave on the same host
>>> (local host). 
>>> but i would like to know the configuration from slony master and slony
>>> slave on two several hosts.
>>>
>>> were will defined the cluster and the tables( i think on the master )?
>>>
>>> were will initialized the connection of slony master and the slony
>>> slave?
>>>
>>> were will started the slon process ( i think on the master for the
>>> master and on the slave for the slave )?
>>>
>>> please help me.
>>>
>>> Thank you very much indeed.
>>>
>>>
>>>
>>>