On Mon, Aug 9, 2010 at 5:33 PM, Tomeh, Husam <HTomeh@corelogic.com> wrote:
> This is useful from a capacity perspective where that can be monitored to alert DBAs when a threshold is reached.
If alerting based on connection counts is your concern, you might get
some benefit from the Bucardo project's check_postgres.pl script. One
of the tests it performs is a backend count. It works with any
monitoring package that speaks NRPE, so you can issue alerts based on
either a percentage of your max_connections setting, or on the actual
backend count, as indicated by the arguments with which it's invoked.
It would also be fairly straightforward to store whatever value the
test returns in a db table, log file, Cacti/other RRD-based tools, or
whatever else strikes your fancy, for historical and trending
analysis.
<http://bucardo.org/check_postgres/>
rls
--
:wq