the postgresql.conf file ... for testing, I used 'set enable_seqscan'
after the server was live to turn on/off, this is just hte default ...
tcpip_socket = true
max_connections = 200 # 1-1024
port = 5434
sort_mem = 4024
shared_buffers = 32768
fsync = false
wal_buffers = 32
debug_pretty_print = true
enable_seqscan = false
On Mon, 11 Feb 2002, Tom Lane wrote:
> I'm trying to work out how the planner arrived at the numbers you're
> showing; the hashjoin cost estimate seems a little lower than I'd
> expect. Are you using nonstandard values for any of the planner cost
> factors? (cpu_operator_cost, etc) How about sort_mem?
>
> regards, tom lane
>