Tom Lane wrote:
> wheel <wheel@wheel.not> writes:
> > I guess the reason is that the pg system database etc are lodged in the
> > hive under \base\, and the system db contains the metadata about the db
> > to be restored?
>
> No, the reason why selective restore doesn't work is that all tables in
> a database cluster depend on the same commit log (pg_clog) to determine
> which rows are valid and which are not. What you were trying to do is
> described in the manual (with a warning not to do it) under
> backup/restore procedures:
> http://www.postgresql.org/docs/8.2/static/backup-file.html
>
> Also, if you would like to know more about the physical layout of the
> database contents, it's described here:
> http://www.postgresql.org/docs/8.2/static/storage.html
This is the contents of /data:
PG_VERSION pg_clog/ pg_multixact/
pg_twophase/ postmaster.opts base/
pg_hba.conf pg_subtrans/ pg_xlog/
postmaster.pid global/ pg_ident.conf
pg_tblspc/ postgresql.conf
None of these are optional for restoring a database. They are all
interconected.
--
Bruce Momjian bruce@momjian.us
EnterpriseDB http://www.enterprisedb.com
+ If your life is a hard drive, Christ can be your backup. +