[check_postgres] query_time returning long queries that don't exist?

[andrew at thepw.com]

Hi,

Occasionally I get these Nagios alerts:

PG QUERY TIME;OK;HARD;2;POSTGRES_QUERY_TIME OK: DB "postgres" longest
query: 0s
PG QUERY TIME;CRITICAL;HARD;2;POSTGRES_QUERY_TIME CRITICAL: DB
"postgres" longest query: 117s PID:8775
PG QUERY TIME;CRITICAL;SOFT;1;POSTGRES_QUERY_TIME CRITICAL: DB
"postgres" longest query: 57s PID:8775

They are running this command on the database server itself:

check_postgres.pl --action=query_time --warning=25 --critical=30

The database is logging every query, but I don't have any queries at
that time that match these times or even close to it.

I yum installed check_postgres:

Installed Packages
Name        : check_postgres
Arch        : noarch
Version     : 2.19.0
Release     : 1.el6
Size        : 533 k
Repo        : installed
From repo   : epel
Summary     : PostgreSQL monitoring script
URL         : http://bucardo.org/wiki/Check_postgres
License     : BSD

and running postgres 8.4:

Installed Packages
Name        : postgresql
Arch        : x86_64
Version     : 8.4.13
Release     : 1.el6_3
Size        : 14 M
Repo        : installed
From repo   : updates
Summary     : PostgreSQL client programs
URL         : http://www.postgresql.org/
License     : PostgreSQL

What could be causing these?

Thanks for your help,
Andrew