Rick Otten <rottenwindf...@gmail.com> writes:
> On Thu, Jan 7, 2016 at 12:51 PM, Scott Rankin <sran...@motus.com> wrote:
>> Winner! Both of those settings were set to always, and as soon as I
>> turned them off, the query times smoothed right out.
> Is this generally true for all PostgreSQL systems on Linux, or only for
> specific use cases?
It's fairly well established that the implementation of transparent
huge pages in Linux kernels from the 2.6-or-so era sucks, and you're
best off turning it off if you care about consistency of performance.
I am not sure whether modern kernels have improved this area.
I think you can get an idea of how big a problem you have by noting
the accumulated runtime of the khugepaged daemon.
(BTW, it would likely be a good thing to collect some current wisdom
in this area and add it to section 17.4 of our docs.)
regards, tom lane
--
Sent via pgsql-performance mailing list (pgsql-performance@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-performance
