David Carpio wrote
> I have made some changes in my postgresql.conf, well, I made two changes
> in this file. the first time, my queries had improved on their execution
> time considerably but in the second change, I seem my queries have not
> improved on the contrary they have come back to be slow or at best, they
> have not changed in its previous improvement.
>
> These are my changes:
>
> + shared_buffers = 4GB.
> + bgwriter_lru_maxpages = 250.
> + synchronous_commit = off.
> + effective_io_concurrency = 3.
> + checkpoint_segments = 64.
> + checkpoint_timeout = 45min
> + logging_collector = on
> + log_min_duration_statement = 500
> + log_temp_files = 0.
>
> my max connections are 150
>
> Please, what would be my error?
>
> Thank you for the tips,
>
> David Carpio
It might increase the likelihood of a meaningful response if you include:
1) The default for each of the parameters
2) The value you used foe each parameter in the situation where performance
improved
Very few people memorized the first and it would be interesting to have the
second for reference.
Also,
Performance questions are hardware/software specific. What are you running
and how are you testing?
You should also do some reading about performance question posting and
performance in general:
https://wiki.postgresql.org/wiki/SlowQueryQuestions
One question is whether you really want to make the default
"synchronous_commit" setting to be "off"? Can you even measure the actual
difference in your specific use-case that turning this off makes or are you
just throwing stuff against the wall that say "this can improve performance"
and hoping things work out for the best? Since it can be turned on/off on a
per-transaction basis you should generally try to only have it off in areas
that are meaningful and were you've acknowledged the corresponding
additional risk specific to that area.
David J.
--
View this message in context:
http://postgresql.1045698.n5.nabble.com/My-changes-in-the-postgresql-conf-does-not-work-tp5762369p5762418.html
Sent from the PostgreSQL - performance mailing list archive at Nabble.com.
