Thorsten Strufe <strufe.pub@googlemail.com> wrote:
> is there any (sane) way to simply get postgre to entirely forget
> about the broken db - and that's it (dumping the other db and
> loading it back might not be the best solution, since it's rather
> large..)?
I think that using pg_dump for the good databases is sane, and the
simplest solution. Another option would normally be one of the
trigger-based replication systems, like Slony (to reduce down time
to a matter of minutes or seconds); but the crippled state of the
cluster may make the hard.
http://wiki.postgresql.org/wiki/Replication,_Clustering,_and_Connection_Pooling
-Kevin