[pgpool-general: 205] Another question about num_init_children and max_pool

[Erwan]

Hi there,

I use a pgpool-II v3.0.5 compiled and installed by my self on a Debian 
Squeeze.
I only use pgpool-II as a loadbalancer between 2 database nodes which 
stay synchronized via an Hot Standby replication.
In my pgpool-II configuration one of this node has a really high weight 
than the other (1000 against 1) to be the prior node to answer.

Everything works well but I have a question about the pool of connections.
In my pgpool-II configuration I have the following settings:
...
num_init_children = 19
max_pool = 4
...

I set it like that because I believed there would be a maximum of 4x19 = 
76 potential connections.
During a test I launched a lot of long queries... and I noted there were 
only 19 connections at the same time and I have to wait for the end of a 
previous queries to launch the next one...
All my connections came from the same host and used the same database 
and user.
I also read in a previous mail on this mailing list 
(http://www.mail-archive.com/pgpool-general@pgfoundry.org/msg03334.html) 
it's normal it's only 19 active connections at a time.
But I don't really understand how I can reach 76 connections if I only 
have 19 active at one time? And in this case, is the pool really useful? 
I just would like there will be a maximum of 76 connections at a time.
According to another mail 
(http://www.sraoss.jp/pipermail/pgpool-general/2011-December/000070.html) I 
will be tempted by setting up my pgpool-II as following:
...
num_init_children = 76
max_pool = 1
...

If you have any clue to help me to better understand this case, it would 
be very nice!
Thanks in advance.
/Erwan