[Pgpool-general] Setting up connections config for load-balancing

[Tatsuo Ishii]

> > I have just committed changes to pgpool-en.html on CVS HEAD,
> > especially about num_init_children. Of course more enhancement are
> > always welcome.
> 
> Can you explain why I'd ever want to use max_pool > 1?  I don't really 
> understand the purpose of it; the docs make it sound like 
> connection-sharing, but it isn't really.

max_pool defines the number of connection cache entries in connection
pool. Each cache entry corresponds to <user name, database>
pair. Pgpool tries to keep the connection cache util number of used
entry in the cache reaches max_pool. If the cache entry exhausts, the
oldest entry is discarded (thus connection to backend is closed) and
new <user name, database> takes place.

So the number of max_pool is higher, chances are larger to reduce the
connection overhead. Of course more max_pool requries more
max_connections.

> > pgpool_status is not a log file, rather to record backend status. I
> > personally use following setting to log:
> 
> Ah, in the documentation it says that it's a log file.

Really? Can you be more specific?

>  Why is there the 
> setting "logdir" in pgpool.conf, then?

Just for historical reason.

> Are you planning on adding a 
> built-in log later?

Yeah, it has been in TODO list for a while. Probably stealing logger
process code from PostgreSQL is the easiest. I just don't have time
until now...
--
Tatsuo Ishii
SRA OSS, Inc. Japan
English: http://www.sraoss.co.jp/index_en.php
Japanese: http://www.sraoss.co.jp