How should I design this, if it's possible?

The high-level requirement is: we need a way to isolate DDL changes and control them more tightly than data changes. We have a source database where we trust devs to alter any & all data they want real-time, but we need to enforce that DDL changes can't go live without a DBA getting involved.

This would meet our needs, if it worked: set up two Slony clusters. Given a path of nodes A -> B -> C, you could have a cluster that replicates A -> B, and another that would replicate B -> C. If you trigger a DDL change on A, it'll stop at B and require access to the slonik on B to trigger the DDL change to C.

I tried to set this up, though, and it didn't work. The initial copy of data would propagate to node C, but any further changes to data on A wouldn't propagate (it'd stop at B).

Does our (admittedly obtuse) requirement make sense? Can you think of a way to meet that requirement with Slony?

Thanks!

- Matt

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.slony.info/pipermail/slony1-general/attachments/20140212/c4e9ec90/attachment.htm