On Sat, 24 Nov 2007 postgres at dac.e4ward.com wrote:

> Anyway... suppose I don't have the choice (which I have) and have to
> work with rails + slony. Could you provide an opinion other than drop
> rails?

You could use rails but not use the rake migrations for schema updates and 
handle them manually.

I've also seen places where before they have to make a DDL change they just 
uninstall slony, make the DDL change using normal DDL on the master then 
reinstall slony and rebuild the slave.  This only works if your dataset is 
small enough that it can be copied to the slave in a 'reasonable' amount of 
time.

If you do go the route of having your migration driver redirect to a shell 
script that calls EXECUTE SCRIPT you might be able to write a program that 
detects any non-replicated tables creates a new set, adds the tables, 
replicates the data and merges the set.  I wouldn't rush ahead and make DDL 
changes in the cluster too transparent or automated, I think building steps 
into the process where DBA has to look at the schema changes and stop to 
think about their impact is a good thing.

You would want to be aware of slony locking behaviors with each of these 
approaches.


Steve

> Thank you.
> -- 
> Diego Algorta Casamayou
> http://www.oboxodo.com - http://diego.algorta.net
> _______________________________________________
> Slony1-general mailing list
> Slony1-general at lists.slony.info
> http://lists.slony.info/mailman/listinfo/slony1-general
>