Re: pg_locks view versus prepared transactions

Поиск
Список
Период
Сортировка
От Tom Lane
Тема Re: pg_locks view versus prepared transactions
Дата
Msg-id 22099.1119298683@sss.pgh.pa.us
обсуждение исходный текст
Ответ на Re: pg_locks view versus prepared transactions  (Alvaro Herrera <alvherre@surnet.cl>)
Ответы Re: pg_locks view versus prepared transactions  (Alvaro Herrera <alvherre@surnet.cl>)
Список pgsql-hackers
Alvaro Herrera <alvherre@surnet.cl> writes:
> On Mon, Jun 20, 2005 at 03:55:45PM -0400, Merlin Moncure wrote:
>> It is interesting to note that systems with stats disabled are unable to
>> get lock owner information in this case (so what?).

> We could make the pg_stat_activity view show information from the
> ProcArray shared struct, when stats are disabled.

pg_stat_activity is not real time, and should not be because its
intended use is to help interpret the also-not-real-time other
statistics.

If people are concerned about this, my inclination would be to add a
separate view, or just add even more columns to pg_locks?

However, all you could get from ProcArray would be the database in
which the backend is running, and maybe the owning user's ID if we
cared to expend the extra space to store it there.  We're certainly
not going to add current_query or any such thing into that array.
        regards, tom lane


В списке pgsql-hackers по дате отправления:

Предыдущее
От: Tom Lane
Дата:
Сообщение: Re: pg_locks view versus prepared transactions
Следующее
От: Josh Berkus
Дата:
Сообщение: Re: Checkpointing problem with new buffer mgr.