Στις Παρασκευή 15 Σεπτέμβριος 2006 17:47, ο/η Tom Lane έγραψε: > Achilleas Mantzios <achill@xxxxxxxxxxxxxxxxxxxxx> writes: > > What considerations should someone take if he is about to just > > cp -r <path_to_some_other_pgsql_data_path>/data . > > .i.e. creating the PGDATA dir *not* following the standard initdb > > procedure > > If you copy the *entire* data tree (data, xlog, clog, any outside > tablespaces), and use exactly the same Postgres executables at both > ends, then it should pretty much Just Work. > Thats what i feared from the begining. (some connection between the data dir and the executable, really, why does the executable need to be exactly the same, if we are talking about minor postgresql upgrade 7.4.12->7.4.13?) The old system is kind of old, and there would be libraries incompatibilities if a was to retain the old executable. So i'll do it the formal way, and pray lotus notes will behave :) Thanx. > > chown -R postgres:postgres ./data > > (new postgres user on the new SUSE might have a different uid/gid than > > the original Debian), > > This is OK. > > > I'd suppose this might work on a test system, but should someone > > rely on it for a critical production system? > > As long as the DBA has practiced beforehand ;-), sure. > > regards, tom lane > > ---------------------------(end of broadcast)--------------------------- > TIP 2: Don't 'kill -9' the postmaster -- Achilleas Mantzios
