Thread: Fastest way to check database's existence

> Is there an even faster way to check for a database's existence that does
> not require establishing a connection?
> (Maybe this question would be more suitable for the pgsql-performance list?)

No. You have to connect to some database in particular to do anything.
That's why the postgres database exists.


--
Regards,
Peter Geoghegan

On 16/10/2010 16:13, Kynn Jones wrote:
> I want to code a Perl function (part of a Perl library) for determining
> the existence of a particular database (in a given host/port).
>
> One way would be to just attempt making a connection to it, trapping any
> errors upon failure (with eval), or discarding the connection upon success.
>
> This approach has the added benefit of also checking the accessibility
> of the database to the user running the code, but for this application
> this added benefit is not necessary.  Checking existence is all that
> matters.
>
> Is there an even faster way to check for a database's existence that
> does not require establishing a connection?

You're going to have to connect no matter what you do, assuming that
you're accessing it from another machine.

An alternative, if it suited your application, would be to maintain a
connection to a database which you know exists, such as template1 or
(better) postgres, and just query pg_database for the existence of the
database you want. If you can keep a connection open for long periods,
I'm sure this would be pretty fast.

Ray.


--
Raymond O'Donnell :: Galway :: Ireland
rod@iol.ie

On Saturday 16 October 2010 8:13:12 am Kynn Jones wrote:
> I want to code a Perl function (part of a Perl library) for determining the
> existence of a particular database (in a given host/port).
>
> One way would be to just attempt making a connection to it, trapping any
> errors upon failure (with eval), or discarding the connection upon success.
>
> This approach has the added benefit of also checking the accessibility of
> the database to the user running the code, but for this application this
> added benefit is not necessary.  Checking existence is all that matters.
>
> Is there an even faster way to check for a database's existence that does
> not require establishing a connection?

Given the restriction of host/port you probably need to connect as stated in
other posts. For completeness though, it is possible to parse
$DATA/global/pg_database for the existence of database in a particular cluster.
Assuming you can map host/port to cluster then it is a possibility.

>
> (Maybe this question would be more suitable for the pgsql-performance
> list?)
>
> TIA!
>
> ~kynn



--
Adrian Klaver
adrian.klaver@gmail.com