Re: [GENERAL] Weird performance difference

On Sat, Oct 21, 2017 at 9:38 AM, Tom Lane wrote: > Also try explicitly ANALYZE'ing the foreign tables. I do not > believe auto-analyze will touch foreign tables ... Autovacuum and autoanalyze only process relations and matviews, discarding the rest when scanning pg_class.

Re: [GENERAL] Weird performance difference

On 2017-10-20 16:38, Tom Lane wrote: Israel Brewster writes: Can you send "explain" (not explain analyze) for the production server? Not for the full query - it only just completed, after 70 minutes or so, and I wasn't running under EXPLAIN ANALYZE. Running with a

Re: [GENERAL] Weird performance difference

Israel Brewster writes: >> Can you send "explain" (not explain analyze) for the production server? > Not for the full query - it only just completed, after 70 minutes or so, and > I wasn't running under EXPLAIN ANALYZE. Running with a shorter date range of > only 7 days,

Re: [GENERAL] Weird performance difference

On Oct 20, 2017, at 3:31 PM, Justin Pryzby wrote: > > On Fri, Oct 20, 2017 at 03:08:26PM -0800, Israel Brewster wrote: >> Summary: the following query takes around 12 seconds on my test machine. On >> my production machine, it's at half an hour and counting. What's going

Re: [GENERAL] Weird performance difference

On Fri, Oct 20, 2017 at 03:08:26PM -0800, Israel Brewster wrote: > Summary: the following query takes around 12 seconds on my test machine. On > my production machine, it's at half an hour and counting. What's going on? > > which, when run on my test server, has this explain analyze output: >

[GENERAL] Weird performance difference

Summary: the following query takes around 12 seconds on my test machine. On my production machine, it's at half an hour and counting. What's going on?Details:As a first stab at getting some data I need, I've developed the following SQL query:SELECT    legfrom,    count(*) as totaldeps,    count(*)