Again with the JPG files?  Please... turn off your HTML.  Thanks.

On Thu, Oct 15, 2009 at 1:30 AM, Lawrence Giam
<lawrenceg at globalitcreations.com> wrote:
>
> at a daily dump of the database for backup purpose and for worst case scenario where all else fails. The slony documentation mention this slony1_extract_schema.sh script which I feel is rather risky to be running on a daily basis and resource consuming on a production server

Does your schema  change on a daily basis?  If so, then you're really
doing something out of the norm, and likely have a flawed design.  I
extract my schema only on changes, which are few and far between.

> 1. Is there a procedure for simplfying the dumping of database without cluster information which can be run on a daily basis?
> 2. Can anyone share how you actually implement daily backup with Slony?
> 3. Lastly how is the restore procedure like?

Just take the dump from the origin, to ensure everything is
consistent.  Do the dump in "c" format.  When you need to restore,
first get a listing of the objects in the dump using pg_restore -l
then edit that listing to eliminate any objects related to slony and
the slony schema, then restore just the remaining objects with
pg_restore -L.

The alternative is to take a data-only dump from either server, then
use your prepared "extracted" schema (up to but not including the
constraints and indexes) then restore to that.  Finally, apply the
remainder of the extracted schema with the indexes and FKs and other
constraints.  The extracted schema file is easy to split since you
just have to look for the first ADD CONSTRAINT and split it in half
right there.