[pgpool-general: 321] Random Connection Failures and processes in status D

[DonMartio]

Hello,

we are evaluating pgpool the first time because we hit a serious 
connection Problem
(Increasing virtual xen maschines leads to increasing max_connections).
Running 2 Postres 9 Instances for failover reasons we thought it would 
be a good idea
to run pgpool in master slave mode.

First tests ran well but we recognized a serious increase of loadavg on 
the pgpool maschine.
The reason is an increasing count of pgpool processes running in status D.
Netstat tells us that they are in CLOSE_WAIT. We can't kill them. We have to
restart pgpool.

What ist going wrong here?
Any hint appreciated

Greetings so far
Martin