On Wed, Sep 12, 2012 at 11:40 PM, Ants Aasma <ants(at)cybertec(dot)at> wrote:
> On Wed, Sep 12, 2012 at 7:00 PM, Jeff Janes <jeff(dot)janes(at)gmail(dot)com> wrote:
>> Regarding the wiki page on reporting slow queries:
>> We currently recommend EXPLAIN ANALYZE over just EXPLAIN. Should we
>> recommend EXPLAIN (ANALYZE, BUFFERS) instead? I know I very often
>> wish I could see that data. I don't think turning buffer accounting
>> on adds much cost over a mere ANALYZE.
> Given the amount of version 8 installs out there the recommendation
> should be qualified with version >9.0. Otherwise a strong +1
>> Also, an additional thing that would be nice for people to report is
>> whether long running queries are CPU bound or IO bound. Should we add
>> that recommendation with links to how to do that in a couple OS, say,
>> Linux and Windows. If so, does anyone know of good links that explain
>> it for those OS?
> I don't have any links for OS level monitoring, but with version 9.2
> track_io_timing would do the job.
I don't know how to advice people on how to use this to obtain
information on a specific query. Would someone else like to take a
stab at explaining that?
In response to
pgsql-performance by date
|Next:||From: Edson Richter||Date: 2012-09-26 20:20:01|
|Subject: Re: [PERFORM] Inaccurate Explain Cost|
|Previous:||From: Shaun Thomas||Date: 2012-09-26 20:03:03|
|Subject: Re: Inaccurate Explain Cost|