Re: Use of sequence rather than index scan for one text column on one instance of a database

Bill Kirtley <bill(at)actbluetech(dot)com> writes:
> For what it's worth:

> select xmin,* from pg_index where indexrelid =
> 'users_key_index'::regclass;
> xmin | indexrelid | indrelid | indnatts | indisunique | indisprimary
> | indisclustered | indisvalid | indcheckxmin | indisready | indkey |
> indclass | indoption | indexprs | indpred
> ------+------------+----------+----------+-------------+--------------
> +----------------+------------+--------------+------------+--------
> +----------+-----------+----------+---------
> 1006 | 15076176 | 17516 | 1 | f | f
> | f | t | f | t | 10
> | 10042 | 0 | |

Uh ... 'users_key_index'? What's that? What would actually be the most
useful is to compare the pg_index entries for the working and
non-working indexes (the ones on email and key).

regards, tom lane