At work we have a 24 cores server, with a load average around 2.5.
I don't know yet if a system which use some unused CPU to minimize the load of a bad query identified early is bad or worse.
Indeed, I don't know if my boss would let me test this at production too, but it could be good to know how things work in "auto-pilot" mode.
2011/12/10 Tomas Vondra
<tv@fuzzy.cz>There's auto_explain contrib module that does exactly what you're asking
for. Anyway, explain analyze is quite expensive - think twice before
enabling that on production server where you already have performance
issues.
Tomas
On 10.12.2011 17:52, Daniel Cristian Cruz wrote:
> Hi all,
>
> I'm trying to figure out some common slow queries running on the server,
> by analyzing the slow queries log.
>
> I found debug_print_parse, debug_print_rewritten, debug_print_plan,
> which are too much verbose and logs all queries.
>
> I was thinking in something like a simple explain analyze just for
> queries logged with log_min_duration_statement with the query too.
>
> Is there a way to configure PostgreSQL to get this kind of information,
> maybe I'm missing something? Is it too hard to hack into sources and do
> it by hand? I never touched PostgreSQL sources.
>
> I'm thinking to write a paper that needs this information for my
> postgraduate course. The focus of my work will be the log data, not
> PostgreSQL itself. If I succeed, maybe it can be a tool to help all of us.
>
> Thank you,
> --
> Daniel Cristian Cruz
> クルズ クリスチアン ダニエル
--
Daniel Cristian Cruz
クルズ クリスチアン ダニエル