Andrew Sullivan ajs at crankycanuck.ca
Thu Oct 18 04:53:17 PDT 2007
On Thu, Oct 18, 2007 at 11:42:40AM +0200, Henry - Zen Search SA wrote:
> > restart from scratch. If you can afford the time to do this, that
> > should work.
> 
> ja, I though of that - just such a mission for a single table though.  I'm
> loath to take that horrible Microsoft approach.

I don't think it's a "Microsoft approach".  What you had was a script
that did something akin to zapping /etc or /usr/bin or such like on a
UNIX box -- some people would conclude that it's easier to reinstall
the OS and then restore from backup than to try to patch up the
running system.

> Muck is the right word.  Slony's (seems) great, but crikey, what a
> complicated system.  Then again, maybe it just seems that way because it's
> so bloody complicated to work with.

Well, it's a very flexible tool with powerful capabilities.  More
powerful systems tend to be more complicated.  That said. . .

> I think slony needs *another* layer on top of slonik - so much of it is
> hit and miss.

. . .slonik was intended to be a bootstrapping user interface, and
Jan always hoped someone would make a _nice_ UI for Slony.  I gather
that (mumble pgadmin?) has done some work in that direction.

> How about this:  how does slony determine the last argument for the
> _db_cluster.logtrigger() trigger?

It has to do with the kinds of data that will be sent.  I think you
don't want to be fooling with this unless you already understand the
C code that underlies the trigger. 

> If I can determine what to stick in there, then I can try recreating that
> trigger manually and see what happens.

I would suggest you not do it that way.  See my other post about how
to refresh this table.  You need to rebuild it from scratch.  You
don't have to drop everything; just that table.

A
-- 
Andrew Sullivan  | ajs at crankycanuck.ca
The fact that technology doesn't work is no bar to success in the marketplace.
		--Philip Greenspun


More information about the Slony1-general mailing list