Le 27/03/2017 à 15:12, Olivier Bernhard a écrit :
> Hi,
> 
> I’m currently trying to solve an issue i have on a slony based replication
> platform (slony 1.94):

I'm not sure this is a version number related to available Slony branches.

> 
> There’s a master postgresql database server (on which master nodes are running)
> and a slave postgresql database server ( on which slave nodes are running).
> On the master server I have the following databases :
> DB1, DB3, DB5, DB7 ….
> On the slave server I have the following database :
> DB2, DB4,DB6, DB8 ….

> Basically, the following replication is expected :
> DB1 -> DB2
> DB3 -> DB4
> DB5 -> DB6
> Etc ….

> 
> However I have noticed I have many processes which are not expected to be here.
> One of these databases is called “ferrandi” and has the same name on the slave
> database server.

You told us just 2 lines upper that DB don't have same name on provider and
subscriber.

So, if I understand correctly, you have two servers and a bunch of databases.
How many sets of replication did you set ?

Could you paste your configuration ?
Cluster, nodes, set...

Maybe by having a look at slony schema in the databases (sl_nodes, sl_set,
sl_tables...)

(...)
>  
> 
> ferrandi database is node 9 while the slave is node 10. So there should be no
> relation between node 9 and nodes 1,2,3,4,5,6,7,8,11,12…20 which are other
> databases. Only node 10 should be working with node 9.
> 
>  
> 
> Checking the configuration in the slony schema, I can see that whatever the
> master database is, the configuration table contain all the nodes, while I
> guess only nodes 9 and 10 should be referenced in this specific replication
> which only involves node 9 and node 10.

I guess you have a single slony cluster, and everything goes into it.

Slony relies on the definition of a cluster, and a network of nodes all able to
communicate with each other.

> 
>  
> 
> Could someone confirm that is node 9 and 10 are only supposed to work together
> (with no other database involved) then I should not have all the nodes declared
> in the configuration ? If so then I guess people who have done this have
> applied a single same configuration file to all the master databases, and this
> may explain why I have all these processes.
> 

Let us see you configuration, but, it may be that everything is configured
within the same cluster, and without any restriction on sl_path.

-- 
Dr. Stéphane Schildknecht

Contact régional PostgreSQL pour l'Europe francophone
Gérant de Loxodata, société de conseil, support et formation

01.79.72.57.75