Speaking as a relative newbie to Slony, that's my understanding as well
but I think the amount of resources saved would be extremely minor any
way you look at it and I'm not sure why you'd want to have something
that needs to be configured at the time of failover when it could have
been done long before. I'd rather have everything pre-configured.

-----Original Message-----
From: slony1-general-bounces at lists.slony.info
[mailto:slony1-general-bounces at lists.slony.info] On Behalf Of Gordon
Shannon
Sent: Monday, September 07, 2009 9:03 AM
To: slony1-general at lists.slony.info
Subject: [Slony1-general] Does every node really need a path to every
othernode?


The documentation at
http://www.slony.info/documentation/slonylistenercosts.html states that 

"It is necessary to have sl_listen entries allowing connection from each
node to every other node. Most will normally not need to be [used] (sic)
very heavily, but it still means that there needs to be n(n-1) paths."

I have been running perfectly fine for several months where this is not
true.  I have a master node A with 4 sets.  Node B replicates all 4 sets
and
there is of course a path between A and B.  Node C replicates 2 sets and
there is a path between A and C.  But there is no path between B and C.


As far as I know, this simply means that I will be restricted if I try
to do
things like FAILOVER or MOVE SET.  I'm thinking if I need to do that, I
can
add the necessary paths at that time.  But not having the extra path
from
B-C all the time saves resources.

In any case, the documentation implies that I MUST have the connection,
and
that does not appear to be the case.  

Thanks,
Gordon

-- 
View this message in context:
http://www.nabble.com/Does-every-node-really-need-a-path-to-every-other-
node--tp25332701p25332701.html
Sent from the Slony-I -- General mailing list archive at Nabble.com.

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