Re: Slow Count-Distinct Query - Mailing list pgsql-performance

From Christopher Jackson
Subject Re: Slow Count-Distinct Query
Date
Msg-id CAN81C19ZLf9C5rJVVY8ijfDLYhxUqC3SBUcdWTPe+ASQJbmZyw@mail.gmail.com
Whole thread Raw
In response to Re: Slow Count-Distinct Query  (Michael Paquier <michael.paquier@gmail.com>)
List pgsql-performance


    Hi Bricklen,

   Thanks again for the feedback.  The concurrent refresh sounds cool.  I just saw the 9.4 release is tentatively scheduled for later this year.  Do you know what people have been doing for view refreshes in the meantime?

   Thanks


On Tue, Apr 1, 2014 at 11:48 PM, Michael Paquier <michael.paquier@gmail.com> wrote:
On Wed, Apr 2, 2014 at 1:22 PM, Christopher Jackson <crjackso@gmail.com> wrote:
>
>     Hi Bricklen,
>
>     Thanks for the feedback.  I'll play around with materialized views.  My
> understanding is they have to be manually triggered for refresh
Yep.

> and there's an exclusive lock on the view while the refresh is taking place.  Is this
> your understanding as well?
Re-yep.

> I'm using PG 9.3.3.  If this is true, I'm
> curious what clever ways people have come up with to mitigate any issues
> with the lock.
Kevin Grittner has implemented REFRESH MATERIALIZED VIEW CONCURRENTLY
in 9.4. A unique index is needed on the materialized view as well to
authorize this concurrent operation. It has the merit to allow SELECT
operations on the matview during the refresh.
--
Michael

pgsql-performance by date:

Previous
From: Michael Paquier
Date:
Subject: Re: Slow Count-Distinct Query
Next
From: Alexey Klyukin
Date:
Subject: Re: Why shared_buffers max is 8GB?