On Tue, Jul 27, 2010 at 15:58, Irene Barg <ibarg@noao.edu> wrote:
> I'm afraid I don't understand how turning on 'log_temp_files' will help?
Well, postgres is probably doing exactly what its told. You just need
to figure out what that is and why. ( and if it is some kind of bug,
we wont be able to do anything about it unless you give us a test case
:-) )
The idea behind log_temp_files is that you should be able to see whats
causing the creation of all those temp files (and so potentially 'fix'
the query/problem). The main problem with log_temp_files is it only
logs when the command is done executing. Of course if it really is
creating tons of files it should take long enough that you should be
able to select * from pg_stat_activity; and pinpoint the problem query
that way. If all else fails you might have better luck with turning
on log_statements and then lurking through the logs when you start to
get lots of tmp files.
> Right now it is set to '-1' which should mean it's disabled, yet it is
> creating 1GB files and NOT cleaning them up:
It wont clean them up until its done using them. If a backend
crashes, it looks like it might orphan some tmp files, at least until
postgres is restarted.
