Re: track_planning causing performance regression

pá 3. 7. 2020 v 13:02 odesílatel Fujii Masao <masao.fujii@oss.nttdata.com>
napsal:

>
>
> On 2020/07/03 16:02, Pavel Stehule wrote:
> >
> >
> > pá 3. 7. 2020 v 8:57 odesílatel Fujii Masao <masao.fujii@oss.nttdata.com
> <mailto:masao.fujii@oss.nttdata.com>> napsal:
> >
> >
> >
> >     On 2020/07/03 13:05, Pavel Stehule wrote:
> >      > Hi
> >      >
> >      > pá 3. 7. 2020 v 4:39 odesílatel Fujii Masao <
> masao.fujii@oss.nttdata.com <mailto:masao.fujii@oss.nttdata.com> <mailto:
> masao.fujii@oss.nttdata.com <mailto:masao.fujii@oss.nttdata.com>>> napsal:
> >      >
> >      >
> >      >
> >      >     On 2020/07/01 7:37, Peter Geoghegan wrote:
> >      >      > On Tue, Jun 30, 2020 at 6:40 AM Fujii Masao <
> masao.fujii@oss.nttdata.com <mailto:masao.fujii@oss.nttdata.com> <mailto:
> masao.fujii@oss.nttdata.com <mailto:masao.fujii@oss.nttdata.com>>> wrote:
> >      >      >> Ants and Andres suggested to replace the spinlock used in
> pgss_store() with
> >      >      >> LWLock. I agreed with them and posted the POC patch doing
> that. But I think
> >      >      >> the patch is an item for v14. The patch may address the
> reported performance
> >      >      >> issue, but may cause other performance issues in other
> workloads. We would
> >      >      >> need to measure how the patch affects the performance in
> various workloads.
> >      >      >> It seems too late to do that at this stage of v13.
> Thought?
> >      >      >
> >      >      > I agree that it's too late for v13.
> >      >
> >      >     Thanks for the comment!
> >      >
> >      >     So I pushed the patch and changed default of track_planning
> to off.
> >      >
> >      >
> >      > Maybe there can be documented so enabling this option can have a
> negative impact on performance.
> >
> >     Yes. What about adding either of the followings into the doc?
> >
> >           Enabling this parameter may incur a noticeable performance
> penalty.
> >
> >     or
> >
> >           Enabling this parameter may incur a noticeable performance
> penalty,
> >           especially when a fewer kinds of queries are executed on many
> >           concurrent connections.
> >
> >
> > This second variant looks perfect for this case.
>
> Ok, so patch attached.
>

+1

Thank you

Pavel


> Regards,
>
> --
> Fujii Masao
> Advanced Computing Technology Center
> Research and Development Headquarters
> NTT DATA CORPORATION
>