Hi!
2009.01.31. 10:13 keltezéssel, Erik Jones írta:
>
> On Jan 30, 2009, at 11:37 AM, durumdara wrote:
>
>> The main viewpoints:
>> - quick (re)connect - because mod_python basically not store the
>> database connections persistently
>
> mod_python is not a database adaptor.
Yes, I know. But because of mod_python (apache) commonly configured to
"1 interpreter/1 request", I need to reconnect with the adapter.
But I tested it, and it is seems to be have enough speed. If not, I must
use some proxy server to keep them alive.
> rewrite the table data and previously existing indexes as well.
> What's more, Postgres allows you to create real tablespaces so that
> you can place individual persistent database objects (databases,
> tables, indexes, and some constraints) on separate storage.
I wanna ask that if I create a database in another tablespace, is the
database name already stored in "main" data storage (like global metadata)?
And If I makes a table or other object into this database, need I to
define the tablespace name, or it is taken from the database
"tablespace" name?
For example:
create db anything tablespace "new2"
These sqls are different?
create table any_table (a int)
or
create table any_table (a int) tablespace "new2".
So: need I define tablespace instruction in every table/object creation,
or not?
And: if I store this database in another drive, and it is unmounted
(because of any reason - like failure, etc), is it causes any problems
with postgresql main databases (for example: service stops, etc). I ask
about "after restart pg service", not online usage.
So if I remove some tablespace path (directory, drive) in the level of
the filesystem, can pg service up to serve main databases the it can access?
Or is it completely die in this problem?
Thanks for your help:
dd