Are there tuning parameters that don't take effect immediately? - Mailing list pgsql-performance

From Michael Nolan
Subject Are there tuning parameters that don't take effect immediately?
Date
Msg-id CAOzAqu+e979Ezq94W5SpCrJ_9iiW7HyxU7eaQFo64rPKhvZS8w@mail.gmail.com
Whole thread Raw
Responses Re: Are there tuning parameters that don't take effect immediately?  ("David G. Johnston" <david.g.johnston@gmail.com>)
Re: Are there tuning parameters that don't take effect immediately?  ("Joshua D. Drake" <jd@commandprompt.com>)
List pgsql-performance
Last night I was doing some tuning on a database  The longest query I was running was taking around 160 seconds.  I didn't see much change in the running time for that query, even after restarting PG.

Today, with roughly the same system load (possibly even a bit heavier load), that query is running about 40 seconds. 

Are there tuning parameters in postgresql.conf that don't take effect right away, even after a restart of PG?  The only thing I can come up that's happened since last night was that we ran the nightly vacuum analyze on that database, but I did not change the statistics target. 

The parameters I was working with were:

effective_cache_size
shared_buffers
temp_buffers
work_mem
maintenance_work_mem

Looking at the free command, I see a lot more memory being used for buffer/cache today.  (Centos 7.)
--
Mike Nolan

pgsql-performance by date:

Previous
From: Xenofon Papadopoulos
Date:
Subject: Re: pg bouncer issue what does sv_used column means
Next
From: "David G. Johnston"
Date:
Subject: Re: Are there tuning parameters that don't take effect immediately?