Re: Query run in 27s with 15.2 vs 37ms with 14.6

Greetings,

* Charles (peacech@gmail.com) wrote:
> Thank you for the hint. I think this is it. random_page_cost was set to 1.1
> in 14.6 (the data directory is located in a ssd). When upgrading to 15.2 I
> thought that ssd random seek time is not that close to sequential seek
> time, so I revert it back to 4.

In the future, you might want to be sure and include 'settings' in your
EXPLAIN options, eg:

EXPLAIN (ANALYZE, BUFFERS, SETTINGS) query ...

so that you see what non-default planner settings are set when the query
is being planned.

Thanks,

Stephen