| From: | Peter Eisentraut <peter(dot)eisentraut(at)2ndquadrant(dot)com> |
|---|---|
| To: | Jim Mlodgenski <jimmy76(at)gmail(dot)com>, PgHacker <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: mat views stats |
| Date: | 2017-02-21 22:22:22 |
| Message-ID: | 09a552a0-9bd7-ee5a-1597-b6c20023a172@2ndquadrant.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On 2/20/17 10:06, Jim Mlodgenski wrote:
> I've come across a number of times where the statistics on materialized
> views become stale producing bad plans. It turns out that autovacuum
> only touches a materialized view when it is first created and ignores it
> on a refresh. When you have a materialized view like yesterdays_sales
> the data in the materialized view turns over every day.
That sounds like a bug.
> Attached is a patch to trigger autovacuum based on a matview refresh
> along with a system view pg_stat_all_matviews to show information more
> meaningful for materialized views.
It might be easier to include materialized views into pg_stat_*_tables.
I think these should be two separate patches. We might want to
backpatch the first one.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Peter Eisentraut | 2017-02-21 22:25:59 | Re: Change in "policy" on dump ordering? |
| Previous Message | Peter Eisentraut | 2017-02-21 22:20:44 | Re: "may be unused" warnings for gcc |