Alberto Dalmaso <dalmaso(at)clesius(dot)it> wrote:
> P.S.: i'm trying with all enable_* to on and pumping to higher
> values from_collapse_limit and join_collapse_limit that I've put to
Tom suggested that you set those numbers higher than the number of
tables joined in the query. I don't think 30 will do that.
> The result is that the query, after an hour of work, goes out of
> memory (SQL State 53200)...
Ouch! Can you provide more details? All information from the
PostgreSQL log about that event would be good. If there's anything
which might be related in the OS logs from around that time, please
include that, too.
Also, with those settings at a high value, try running just an EXPLAIN
(no ANALYZE) of the query, to see how long that takes, and whether you
have a memory issue during the planning phase. (You can use \timing
in psql to get a report of the run time of the EXPLAIN.)
In response to
pgsql-performance by date
|Next:||From: Hartman, Matthew||Date: 2009-06-17 14:58:50|
|Subject: Re: Speeding up a query.|
|Previous:||From: Kevin Grittner||Date: 2009-06-17 14:48:15|
|Subject: Re: performance with query|