Thanks James!
I have some ideas on how implement this using client-provided aggregate
specifiers (think StatsD). I'll check PGXN for anything similar, and if I
don't find anything, will consider engaging pgsql-hackers@ per
https://wiki.postgresql.org/wiki/So,_you_want_to_be_a_developer%3F
Hi Jacob,
This is the same problem as being able to monitor the total number of
processed queries, or the total query processing time.
The only solution I'm aware of is to set pg_stat_statements.max higher than
the distinct number of normalized queries you expect.
As soon as pg_stat_statements
When reviewing execution statistics, I am frequently interested in the
behavior of "classes" of queries, rather than individual queries, for
example queries which
- Contain a join
- Touch a specific column
- Use POSIX regular expressions
AFAIK this sort of summary/rollup information