Joe, 

I see a few issues with what you are proposing.
1. When you restore the data the next xid on the target system might be lower than the xid's used on the source system.  This will break the logic used to connect SYNC events in sl_event with the proper rows in sl_log.  You could get around this by setting the xid wraparound counter on the target system to be higher than the wraparound counter on the source.

2. "All nodes in sync" is a bit of a loaded statement.  Do you mean "all data has been replicated" or all sync events have been confirmed on the remote node.  In order to get to a state where a sync event has been confirmed on a remote node but the next sync event has not yet been generated you will need to do something to prevent the next sync event from being generated.  You can do that by changing the sync interval too something much higher(a slon.conf) option

I think Chris's ideas will be easier to get working in practice, but otherwise I think you could do a backup as you describe if you address the above two issues.

Steve
 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.slony.info/pipermail/slony1-general/attachments/20130206/fbebc8da/attachment.htm