Re: tablespace and sequences?

Поиск
Список
Период
Сортировка
От Fabien COELHO
Тема Re: tablespace and sequences?
Дата
Msg-id Pine.LNX.4.58.0408181614420.31684@sablons.cri.ensmp.fr
обсуждение исходный текст
Ответ на Re: tablespace and sequences?  (Robert Treat <xzilla@users.sourceforge.net>)
Список pgsql-hackers
Dear Robert,

> Chris, help me understand this will you?

I'm not Chris, but it looks like Robert may eventually share my concerns,
so I'm happy not to be alone on this one ;-)

> On my production system I have a few very large tables I want to move
> into their own tablespace so I can but them a a very large disk, and a
> couple frequently updated tables I would like to move into their own
> tablespace so i can put them on their own (small, raid oriented) disk.
> I need to do all this from a physical side of things for performance and
> administration in production, but when I create test databases for
> developers/testing, I don't want to have to recreate the same physical
> layout on every system....

What you describe is basically the reason why I'm advocating, quite
unsuccessfully at the time, that pg_dump/pg_restore should deal with
tablespace in some careful and appropriate manner even in coming 8.0.

> it sounds like you are saying that is the case...

It is indeed the case and the reason for my query about the todo item. The
current status is that you cannot restore a dump if tablespaces where used
if the same tablespaces do not exist in the target system. So it is fine
if you want to restore on the same system, but not on another one. You
would have to create them artificially or to edit them out of the script
if you want a transfer on a different system.

Have a nice day,

-- 
Fabien Coelho - coelho@cri.ensmp.fr


В списке pgsql-hackers по дате отправления:

Предыдущее
От: Robert Treat
Дата:
Сообщение: Re: tablespace and sequences?
Следующее
От: "Magnus Hagander"
Дата:
Сообщение: Re: initdb failure on win32