I guess it depends on what sort of infrastructure you are running.
There are a few ways to backup
http://www.postgresql.org/docs/current/static/backup.html
and depending on your IO load on the DB, how you should backup varies a lot, and if your infrastructure is virtual - it varies again.
Background on the DB stuff:
* Lucene re-index process is a lot of read operations.
* Being Harvested - same thing.
* Services harvest operations - a handful of writes - importance? Not really - you can always re-harvest
* Geonetwork harvest a slave - many transactions.
We use dumps but our data is small (10k+ records) and have not had any consistency issues.
If you have space (and free IO) - I would move to Continuous Archiving - postgres 9.x is pretty good with that.
Regards,
Terry
From: Victor Sinceac [mailto:victor.sinceac@anonymised.com]
Sent: Friday, 16 November 2012 4:14 AM
To: geonetwork-devel@lists.sourceforge.net; geonetwork-users@anonymised.comrge.net
Subject: [GeoNetwork-users] Recipes about backing-up GeoNetwork databases
Hi,
Does someone know if there are known issues for backing up GeoNetwork's database (PostgreSQL+PostGIS) on production installations, on regular basis?
Did someone try to see what happens if db backup process starts while GeoNetwork harvests/is harvested/builds or rebuilds Lucene indexes?
Is there any recipe about this, for large catalogs?
Many thanks,
Victor
--
Victor Sinceac: victor.sinceac@anonymised.com<mailto:victor.sinceac@anonymised.com>, vsinceac@anonymised.com<mailto:vsinceac@anonymised.com>
Address: 24, villa Auguste Blanqui, 75013 Paris, France
Phone: +33 9 5277 0042 / +33 6 9507 0434