Re: [HACKERS] BUG: pg_stat_statements query normalization issueswith combined queries
| От | Fabien COELHO |
|---|---|
| Тема | Re: [HACKERS] BUG: pg_stat_statements query normalization issueswith combined queries |
| Дата | |
| Msg-id | alpine.DEB.2.20.1612201049360.29821@lancre обсуждение исходный текст |
| Ответ на | Re: [HACKERS] BUG: pg_stat_statements query normalization issueswith combined queries (Fabien COELHO <coelho@cri.ensmp.fr>) |
| Ответы |
Re: [HACKERS] BUG: pg_stat_statements query normalization issueswith combined queries
|
| Список | pgsql-hackers |
>> I was expecting the 2 combined queries either to be separated in individual
>> queries "SELECT data FROM Stuff WHERE id = ?" or in one large queries with
>> three ?, but not the above result...
>
> Oops, I forgot the attachement, see repeat script on 9.6.1
After some quick investigation, I concluded that the issue is that the
whole combined query string is used instead of the part refering to the
actual query being processed.
As a result:
- the full un-normalized string is used for both BEGIN & COMMIT => n shared entries, one for each actual query,
begin& commit are mixed together
- the fist-part only normalized string is used for each SELECT => 1 shared entry with "query" is the first partially
normalied encountered combined query
--
Fabien.
В списке pgsql-hackers по дате отправления: