On 10/24/2012 04:58 PM, Christopher Browne wrote:
> Arguably, the failure to restore the triggers isn't a "real" error, if
> you squint at things in a particular way.
> 
> Alternatively, I suppose you could do a separate dump of *just* the
> Slony schema.  That has *some* risk of locking conflict, but it should
> be an entirely smaller risk, as it's not encompassing the likely-large
> backup of your application's schema and data.  Dump the Slony schema,
> and load it first, and that would allow the
> schema-excluding-Slony-bits to work.
> 
> But my preference would be to squint at things sideways and say, "It's
> OK that those triggers didn't load in."

Thanks for the response (and Steve's as well). We are scripting the
restore and therefore prefer to treat all errors as a failure.
Unfortunately if we "allow" the restore trigger errors it will make it
difficult to disallow other errors. But I guess it is something to consider.

And in fact we are already dumping and restoring the slony schema
separately first, and encountered the described deadlock which caused
pg_dump to fail, which prompted the question in the first place.

Joe

-- 
Joe Conway
credativ LLC: http://www.credativ.us
Linux, PostgreSQL, and general Open Source
Training, Service, Consulting, & 24x7 Support