I picked 260 because 256 are the max clients in the apache configuration
file.
My problem is that every apache request need to make a connection to the
database, so if I have all the 256 apache processes working, I need at
least 256 pg_pool processes.
However, with the pg_pgpool installed in each machine with 260 preforked
clients, the test ended quite well. The postgresql server didn't loaded
two much and errors like "There is already a transaction in progress"
disappeared (I think that was a problem with the apache persistent
connections).
I encountered also other problems that I must to check. I think they are
caused by my application.
As soon as I have some results, I'll let you know.
Thank you,
Denis
Jeff Trout wrote:
>
> On Jul 27, 2005, at 10:46 AM, denis@edistar.com wrote:
>
>> I'm now testing with pg_pool installed on each apache frontend with
>> 260 pg_pool preforked clients in each machine.
>
>
> Why did you pick 260?
>
> You don't need a 1:1 ratio. That is the point of the pool. Those
> connections are "shared". Chances are extremely high that all your
> apache clients are not issuing queries at the same exact time so your
> queries end up getting funnelled into those X connections.
>
> I ran with 32 kids on pg_pool and 350 apache processes. never had a
> problem.
>
> --
> Jeff Trout <jeff@jefftrout.com>
> http://www.jefftrout.com/
> http://www.stuarthamm.net/
>
>
