Em 06-11-2012 15:36, Tom Lane escreveu:
> Merlin Moncure<mmoncure@gmail.com> writes:
>> Problem could also be no statistics (run ANALYZE to test) or some
>> other configuration problem (like index locale), or a bona fide
>> regression.
> I'm wondering about join_collapse_limit in particular --- if that wasn't
> cranked up in the 9.1 installation, it would be pure luck if you got a
> good query plan for an example like this.
I tried increasing it from 8 to 25 and it didn't make any difference.
> Maybe that and/or other
> parameter settings didn't get transposed to the 9.2 installation.
diff /etc/postgresql/9.[12]/main/postgresql.conf
41c41
< data_directory = '/var/lib/postgresql/9.1/main' # use
data in another directory
---
> data_directory = '/var/lib/postgresql/9.2/main' # use
data in another directory
43c43
< hba_file = '/etc/postgresql/9.1/main/pg_hba.conf' # host-based
authentication file
---
> hba_file = '/etc/postgresql/9.2/main/pg_hba.conf' # host-based
authentication file
45c45
< ident_file = '/etc/postgresql/9.1/main/pg_ident.conf' # ident
configuration file
---
> ident_file = '/etc/postgresql/9.2/main/pg_ident.conf' # ident
configuration file
49c49
< external_pid_file = '/var/run/postgresql/9.1-main.pid'
# write an extra PID file
---
> external_pid_file = '/var/run/postgresql/9.2-main.pid'
# write an extra PID file
63c63
< port = 5433 # (change requires restart)
---
> port = 5432 # (change requires restart)
556a557,558
> ssl_cert_file = '/etc/ssl/certs/ssl-cert-snakeoil.pem'
> ssl_key_file = '/etc/ssl/private/ssl-cert-snakeoil.key'
Any other idea?