Re: Slow statement using parallelism after 9.6>11 upgrade

Поиск
Список
Период
Сортировка
Искать
От
Arnaud L.
Тема
Re: Slow statement using parallelism after 9.6>11 upgrade
Дата
Msg-id
ffde859b-95f5-5ac9-7ceb-f24717b3b862@codata.eu
Ответ на
Список
Дерево обсуждения
Slow statement using parallelism after 9.6>11 upgrade "Arnaud L." <arnaud.listes@codata.eu>
Re: Slow statement using parallelism after 9.6>11 upgrade Tom Lane <tgl@sss.pgh.pa.us>
Re: Slow statement using parallelism after 9.6>11 upgrade "Arnaud L." <arnaud.listes@codata.eu>
Re: Slow statement using parallelism after 9.6>11 upgrade "Arnaud L." <arnaud.listes@codata.eu>
Re: Slow statement using parallelism after 9.6>11 upgrade "Arnaud L." <arnaud.listes@codata.eu>
Re: Slow statement using parallelism after 9.6>11 upgrade "Arnaud L." <arnaud.listes@codata.eu>
Re: Slow statement using parallelism after 9.6>11 upgrade Paul Ramsey <pramsey@cleverelephant.ca>
Re: Slow statement using parallelism after 9.6>11 upgrade "Arnaud L." <arnaud.listes@codata.eu>
Bad estimates on GIN bigint[] index "Arnaud L." <arnaud.listes@codata.eu>
Re: Slow statement using parallelism after 9.6>11 upgrade Imre Samu <pella.samu@gmail.com>
Re: Slow statement using parallelism after 9.6>11 upgrade "Arnaud L." <arnaud.listes@codata.eu>
Le 03/09/2019 à 17:02, Imre Samu a écrit :
>  > We have upgraded our database from 9.6 to 11
> 
> This is now the latest PG ?   PG11.5?

Yes, PostgreSQL 11.5, compiled by Visual C++ build 1914, 64-bit.


> - the osm2pgsql has an own parallelizations ...  ( osm2pgsql  
> --number-processes .. )
>     so be careful to add more parallelisation to the PG side with the 
> same osm2pgsql parameters !   ( check the memory usages! )
> - check the benchmarks and the tunnings: 
> https://wiki.openstreetmap.org/wiki/Osm2pgsql/benchmarks
> - you can ask help on the  : "osm dev mail list" (="dev OpenStreetMap 
> developer discusssion" https://lists.openstreetmap.org/listinfo/dev )


Yes, I know. It as been fine tuned over some time now to fit our 
server's capacity.
Disabling parallelisme has allowed the update to run in ~5000s, which is 
consistent with what we could witness before the upgrade.

I had started writing an issue on the osm2pgsql github before realizing 
it was probably a PG-only issue.


Cheers
--
Arnaud



В списке pgsql-general по дате отправления
От: Arnaud L.
Дата:
От: Jehan-Guillaume (ioguix) de Rorthais
Дата:
FAQ