Re: [Gnumed-devel] Backup "database" vs backup "data"

[Karsten Hilbert]

On Fri, May 27, 2011 at 10:10:52PM -0700, Jim Busser wrote:

> >From the inline script documentation,
> 
>       gm-backup_database.sh creates an uncompressed, plain
>   text (SQL) backup # of the database schema, data, and roles
>   which can be used to # restore a GNUmed database from
>   scratch with psql.
> 
> whereas
> 
>       gm-backup_data.sh creates a backup of the data of a
>   GNUmed # database (including neither roles nor the schema)
> 
> which got me to thinking… while I can understand that
> people will not need to back up the schema daily –

Backing up roles and schema along with the data is
*entirely* irrelevant in terms of backup size.

The data-only script is merely intended for transferring
data into an existing database created by other means.

We, as GNUmed, strongly recommend using the full backup, not
the data-only one unless the user knows why they want the
data-only solution.

> as the
> schema may not change more often than once every few months
> – how often will the roles change?

Rarely.

> Are we talking about
> 
>               postgres roles AND gnumed users

PostgreSQL roles. GNUmed users are data for what PostgreSQL
knows.

> which means that any (every) time a new doctor or
> secretary is enrolled as a user, it would be important to
> run a full backup (not "just" the "data")

See above.

Karsten
-- 
GPG key ID E4071346 @ gpg-keyserver.de
E167 67FD A291 2BEA 73BD  4537 78B9 A9F9 E407 1346