Re: pgsql 10: hash indexes testing

On Fri, Aug 11, 2017 at 07:33:51AM +0530, Amit Kapila wrote:
> On Thu, Aug 10, 2017 at 4:11 PM, AP <ap(at)zip(dot)com(dot)au> wrote:
> > mdstash=# select * from pgstathashindex('link_datum_id_idx');
> > version | bucket_pages | overflow_pages | bitmap_pages | unused_pages | live_items | dead_items | free_percent
> > ---------+--------------+----------------+--------------+--------------+------------+------------+------------------
> > 4 | 12391325 | 5148912 | 158 | 191643 | 4560007478 | 0 | 1894.29056075982
> > (1 row)
>
> The free_percent calculation seems to be wrong. Can you please once
> try after recent commit 0b7ba3d6474b8f58e74dba548886df3250805cdf? I
> feel this should be fixed by that commit.

Sorry I couldn't get to help you debugging this myself. Work got annoying. :/

That said, I think that this is the first time that I've seen the value be
under 100:

mdstash=# select * from pgstathashindex('link_datum_id_idx');
version | bucket_pages | overflow_pages | bitmap_pages | unused_pages | live_items | dead_items | free_percent
---------+--------------+----------------+--------------+--------------+------------+------------+------------------
4 | 22957200 | 9272698 | 283 | 2208624 | 8448300552 | 0 | 39.8146658879555
(1 row)

Time: 2882974.635 ms (48:02.975)

The index is still functioning, too, with far more data than I've ever had
in the table in the past and well beyond the point where it would previously
die. Happy days. :)

AP