From: | Sergey Burladyan <eshkinkot(at)gmail(dot)com> |
---|---|
To: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
Cc: | Jeff Janes <jeff(dot)janes(at)gmail(dot)com>, "pgsql-performance\(at)postgresql(dot)org" <pgsql-performance(at)postgresql(dot)org> |
Subject: | Re: Looks like merge join planning time is too big, 55 seconds |
Date: | 2013-08-02 16:20:22 |
Message-ID: | 878v0kuks9.fsf@seb.koffice.internal |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-performance |
Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> writes:
> Jeff Janes <jeff(dot)janes(at)gmail(dot)com> writes:
> > On Thu, Aug 1, 2013 at 5:16 PM, Sergey Burladyan <eshkinkot(at)gmail(dot)com> wrote:
> >> If I not mistaken, may be two code paths like this here:
> >> (1) mergejoinscansel -> scalarineqsel-> ineq_histogram_selectivity -> get_actual_variable_range -> index_getnext
> >> (2) scalargtsel -> scalarineqsel -> ineq_histogram_selectivity -> get_actual_variable_range -> index_getnext
>
> > Yeah, I think you are correct.
>
> mergejoinscansel does *not* call scalarineqsel, nor get_actual_variable_range.
> It calls get_variable_range, which only looks at the pg_statistic
> entries.
Hmm, I speak about 9.2.2 but in current HEAD this call still exist,
please see: http://doxygen.postgresql.org/selfuncs_8c_source.html#l02976
> I think we need to see the actual stack traces, not incomplete versions.
> It's possible that the situation here involves bloat in pg_statistic, but
> we're just leaping to conclusions if we assume that that's where the index
> fetches are occurring.
I found debug symbols and send stack trace to mail list, but it blocked
by size, try again with zip
Attachment | Content-Type | Size |
---|---|---|
gdb.log.gz | application/x-gzip | 3.3 KB |
From | Date | Subject | |
---|---|---|---|
Next Message | Alvaro Herrera | 2013-08-02 16:23:44 | Re: Looks like merge join planning time is too big, 55 seconds |
Previous Message | Tom Lane | 2013-08-02 15:50:48 | Re: Looks like merge join planning time is too big, 55 seconds |