Thread: restoring an old database to a new instance -- possible ?
Using PostSQL 7.4 on a linux server, we have one postgres instance with two databases, "gex_runtime" and "mq_geoloc." Wehad a mishap in which we ran out of disk space. The failure occured in writing to the gex_runtime database. Then an unfortunate slip of an admin's keyboard led to the deletion of the gex_runtime data itself under the $PGDATA directory.As far as I can tell the mq_geoloc database files (about 3 gigs worth) were undamaged. We moved the mq_geoloc files to a safe location, and now have taken a copy of the runtime database from another server andrecreated our server as far as the runtime database goes. My question is whether there is anyway of retrieving the old geoloc database ? Or is my reading of these files total out-of-linewith reality ? If not it's not a crisis -- just a matter of time rebuilding table and reindexing it, but if there's a faster way I'd bewilling to try it. Thanks for your time, Greg Williamson DBA GlobeXplorer LLC
"Gregory S. Williamson" <gsw@globexplorer.com> writes: > Using PostSQL 7.4 on a linux server, we have one postgres instance with two databases, "gex_runtime" and "mq_geoloc." Wehad a mishap in which we ran out of disk space. The failure occured in writing to the gex_runtime database. > Then an unfortunate slip of an admin's keyboard led to the deletion of the gex_runtime data itself under the $PGDATA directory.As far as I can tell the mq_geoloc database files (about 3 gigs worth) were undamaged. > We moved the mq_geoloc files to a safe location, and now have taken a copy of the runtime database from another serverand recreated our server as far as the runtime database goes. > My question is whether there is anyway of retrieving the old geoloc database ? Or is my reading of these files total out-of-linewith reality ? If you saved the *whole* $PGDATA directory tree including pg_clog and so on, then there's nothing at all wrong with the mq_geoloc database. Just don't try to connect to gex_runtime. I'd fire up a postmaster and pg_dump mq_geoloc, then reload that data into your new installation. regards, tom lane
Tom -- Thanks for suggestion. We'll give it a try. Greg W. -----Original Message----- From: Tom Lane [mailto:tgl@sss.pgh.pa.us] Sent: Thu 8/11/2005 7:07 AM To: Gregory S. Williamson Cc: pgsql-admin@postgresql.org Subject: Re: [ADMIN] restoring an old database to a new instance -- possible ? "Gregory S. Williamson" <gsw@globexplorer.com> writes: > Using PostSQL 7.4 on a linux server, we have one postgres instance with two databases, "gex_runtime" and "mq_geoloc." Wehad a mishap in which we ran out of disk space. The failure occured in writing to the gex_runtime database. > Then an unfortunate slip of an admin's keyboard led to the deletion of the gex_runtime data itself under the $PGDATA directory.As far as I can tell the mq_geoloc database files (about 3 gigs worth) were undamaged. > We moved the mq_geoloc files to a safe location, and now have taken a copy of the runtime database from another serverand recreated our server as far as the runtime database goes. > My question is whether there is anyway of retrieving the old geoloc database ? Or is my reading of these files total out-of-linewith reality ? If you saved the *whole* $PGDATA directory tree including pg_clog and so on, then there's nothing at all wrong with the mq_geoloc database. Just don't try to connect to gex_runtime. I'd fire up a postmaster and pg_dump mq_geoloc, then reload that data into your new installation. regards, tom lane !DSPAM:42fb5baa311789077918311!