> It has now been decided to try upgrading to 9.4 as that is the minimum to > support Django 1.11 (which we are trying to upgrade a backend service to). > The hope is whatever feature we have not configured properly in 9.6 is not > there in 9.4. It's entirely possible whatever is causing your performance issue is caused by the migration, rather than anything inherently different in 9.6. The best test for that is setting another 9.3 server up, restoring a backup, and testing there. If that is very different than what you are getting on 9.6 then it's something which changed in Postgres, if not it's just bad stats.
I do think that it's probably better to fix your query rather than choosing to upgrade to 9.4 rather than 9.6. You have a crazy amount of your query time spent in a single node. That plan is not good. If that's the only query giving you trouble, work on optimizing it. Just my $0.02 -Adam -- Sent via pgsql-performance mailing list (email@example.com) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-performance