From: | Jeff Janes <jeff(dot)janes(at)gmail(dot)com> |
---|---|
To: | "pgsql-performance(at)postgresql(dot)org" <pgsql-performance(at)postgresql(dot)org> |
Subject: | pg_stat_statements with fetch |
Date: | 2017-05-19 23:04:36 |
Message-ID: | CAMkU=1wW5vX5+PCVAPrmHH6589h-1dLKYqwrwCp18obpTZFUbA@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-performance |
I'm spoiled by using pg_stat_statements to find the hotspot queries which
could use some attention.
But with some recent work, all of the hotspots are of the form "FETCH 1000
FROM c3". The vast majority of the queries return less than 1000 rows, so
only one fetch is issued per execution.
Is there an automated way to trace these back to the parent query, without
having to strong-arm the driving application into changing its cursor-using
ways?
pg_stat_statements v1.4 and postgresql v9.6 (or 10beta1, if it makes a
difference)
Sometimes you can catch the DECLARE also being in pg_stat_statements, but
it is not a sure thing and there is some risk the name got freed and reused.
log_min_duration_statement has the same issue.
Cheers,
Jeff
From | Date | Subject | |
---|---|---|---|
Next Message | Zac Goldstein | 2017-05-20 23:33:16 | Index not being used on composite type for particular query |
Previous Message | Tom Lane | 2017-05-19 16:09:58 | Re: GIN index not used if created in the same transaction as query |