Brian,

The tutorial I found for Slony on Windows has you use a bunch of different
files to create the cluster.  You are correct, I have three database servers
each with two databases.  Both master databases are on the same server & am
I trying to use two replication groups to replicate the data.  I'm not sure
what I did Monday/Tuesday morning but I am now to the point where I can't
even replicate a single table.  Seems like I take 1 step forward on this and
then take 10 steps backwords.

The following is included in every file I pass to slonik when building the
cluster.

cluster name = sr_routing_cluster;
node 1 admin conninfo = 'service=sr_routing_01-slonik';
node 2 admin conninfo = 'service=sr_routing_02-slonik';
node 3 admin conninfo = 'service=sr_routing_03-slonik';

To initialize the cluster, I use a file containing
include <preamble.sk>;
init cluster (id=1,comment='hostname=172.16.6.70');

Adding the nodes
include <preamble.sk>;
store node (id=2, comment='hostname=172.16.6.71', event node=1);
store node (id=3, comment='hostname=172.16.6.72', event node=1);

Adding the paths
include <preamble.sk>;
store path (server=1, client=2, conninfo='service=sr_routing_02-slonik');
store path (server=1, client=3, conninfo='service=sr_routing_03-slonik');
store path (server=2, client=1, conninfo='service=sr_routing_01-slonik');
store path (server=3, client=1, conninfo='service=sr_routing_01-slonik');

Config file for the slony service
cluster_name='sr_routing_cluster'
conn_info='service=sr_routing_01-slonik'

Build the set
include <preamble.sk>;
create set (id=1, origin=1, comment='Spatial Router GIS Data');

Add the table(s)
include <preamble.sk>;
set add table(set id=1, origin=1, id=1, fully qualified name='sde.routing',
comment='Routing', key='routing_pk');

And finally subscribe the slaves
include <preamble.sk>;
subscribe set (id=1, provider=1, receiver=2, forward=yes);
sync(id=1);
wait for event (origin=1, confirmed=2, wait on=1, timeout=30000);
subscribe set (id=1, provider=1, receiver=3, forward=yes);
sync(id=1);
wait for event (origin=1, confirmed=3, wait on=1, timeout=30000);

-----Original Message-----
From: Brian Fehrle [mailto:brianf at consistentstate.com] 
Sent: Wednesday, June 16, 2010 18:41
Cc: Shaun McCloud; slony1-general at lists.slony.info
Subject: Re: [Slony1-general] Replicate multiple databases in Windows

Shaun,

 From looking at your config file, I am assuming you have this type of 
setup. Three physical postgres clusters, 6 total databases (two sets of 
3, where one database of each set is a master, the other two are slaves).

--------------pg Cluster A--------------  <-- slony master 172.16.6.70
----database X---- ----database Y----
------table D------- ------- table H------

--------------pg Cluster B--------------  <-- slony master 172.16.6.71
----database X---- ----database Y----
------table D------- ------- table H------

--------------pg Cluster C--------------  <-- slony master 172.16.6.72
----database X---- ----database Y----
------table D------- ------- table H------

----Replication set 1----
----------table D ---------

----Replication set 2----
----------table H ---------

I didn't see anywhere in the config file where you are defining a slony 
cluster, I don't have any familiarity with slony on windows, so not sure 
how this is handled. Are you trying to get all six nodes running in the 
same slony cluster? If so, you'll need to have two separate "master" or 
"origin" nodes on Cluster A. then for each of the x.71 and x.72 
clusters, you will need two slony nodes set up as subscribers for their 
respective replication sets.

---- Origin nodes ----
Database 'X' on Cluster A will be the origin node for 'replication set 1'
Database 'Y' on Cluster A will be the origin node for 'replication set 2'

---- Subscriber Nodes ----
Database 'X' on Cluster B will be subscribed to 'replication set 1'
Database 'Y' on Cluster B will be subscribed to 'replication set 2'

Database 'X' on Cluster C will be subscribed to 'replication set 1'
Database 'Y' on Cluster C will be subscribed to 'replication set 2'

Another (and probably easier) option is to have two separate slony 
clusters set up. With this, each slony cluster will only need to be 
connected to three nodes (you will still need 6 nodes and a 
service/daemon for each of them)

--- slony cluster sr_routing ---
pg Cluster A - Database X - replication set 1 <-- Origin node
pg Cluster B - Database X - replication set 1 <-- Subscriber node
pg Cluster C - Database X - replication set 1 <-- Subscriber node

--- slony cluster sr_alt_routing ---
pg Cluster A - Database Y - replication set 1 <-- Origin node
pg Cluster B - Database Y - replication set 1 <-- Subscriber node
pg Cluster C - Database Y - replication set 1 <-- Subscriber node

So I'm not quite sure which of these you're attempting to get set up, 
but if the error is something along the lines of asking "if I am using 
the correct database when trying to subscribe the slaves to the first 
replication set", I'd say to make sure that you are only subscribing the 
correct slaves to the first replication set. Since you will have two 
separate replication sets, one for each table that live on separate 
databases, you won't want to subscribe Database 'Y' on pg Cluster C to 
the replication set that Database 'X' on pg Cluster A is the origin node 
for.


I hope I got my assumptions right, as well as my conclusions. Any more 
information you can provide us will help out.

- Brian Fehrle

Steve Singer wrote:
> Shaun McCloud wrote:
>   
>> Hello,
>>
>>  
>>
>> I am trying to replicate two databases in Windows (Web Server 2008R2 to 
>> be exact) using Slony-I 1.2.15 & PostgreSQL 8.3 (my hands are tied on 
>> the PostgreSQL version for right now).  The guide I found to go by is 
>> from the kind folks at EnterpriseDB and works great for one database.  
>> If I setup separate services for each database in my pg_service.conf 
>> file I get event viewer errors asking if I am using the correct database 
>> when trying to subscribe the slaves to the first replication set.  I 
>> know it would be easier to put the two tables (yes, that's all I am 
>> replicating total) into the same database, but if I do that then we have 
>> to modify our customers data and we would like to avoid that if at all 
>> possible.  Am I just trying to do something that isn't possible?
>>     
>
> I should say that Slony under Win32 doesn't seem to get a lot of use, so 
>    I can't rule out their being unreported problems under win32.
>
> In theory you should be able to have both databases on the same server. 
>   What service names are your slons using?  sr_routing_01-slonik and 
> sr_routing_02-slonik are both pointing at the same database, is this 
> what you intended?  I see lots of entries in that conf file but you only 
> have 2 databases (that you have told us about) so I would only expect 
> you to need to slon services.
>
>
>   


>   
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> Slony1-general mailing list
>> Slony1-general at lists.slony.info
>> http://lists.slony.info/mailman/listinfo/slony1-general
>>     
>
>
>