Though some Linux distributions package the check_postgres utility for easy installation, the versions that are included are usually very old. We recommend that you obtain a copy of the latest check_postgres source code. At the time of writing this book, the latest version is 2.21.0, released on September 24, 2013. Obtain the latest copy of the check_postgres source code from http://bucardo.org/wiki/Check_postgres.

As we want to use Nagios to execute the check_postgres, please follow the steps in the Configuring Nagios to monitor a database host recipe to produce a working installation with a basic database host configuration. We will be making further modifications to the db_conf.cfg file introduced there.

Install check_postgres by following these steps:

As the postgres user on a PostgreSQL server, try using these commands to obtain database information:

Integrate check_postgres.pl into Nagios by following these steps:

This recipe comes in three parts because we're doing three distinctly different things. Installing check_postgres itself is actually very easy. The entirety of the utility is contained within a single file, so we can simply move check_postgres.pl to a suitable location in our PATH environment setting. However, we suggest that you use the standard installation process as we did.

Installing /usr/local/bin/check_postgres.pl

Next, we try a couple of basic commands to ensure that check_postgres works. The first command makes use of the database_size action and alerts us if our database is larger than the warning (-w) or critical (-c) thresholds that we set. The table_size action performs a similar task but applies the thresholds to every table in the database. By default, check_postgres connects to the postgres database, so we placed a large table there to trigger a critical alert. The output is very large as it lists every table, but it should begin like this:

POSTGRES_TABLE_SIZE CRITICAL: DB "postgres" (host:192.168.56.10) largest table is "public.bigtable": 35 MB

As we have verified that the check works, we want Nagios to invoke it instead. This removes the need to create ad hoc invocations and allows us to search for large tables on all the database servers that Nagios is monitoring.

We will start the process by adding a command to Nagios in the db_conf.cfg file we created for our single test server. Remember where check_postgres.pl was installed, because we need to specify the full path to the script, just in case it's not part of the standard PATH environment. We will set the first argument to set the action we want to perform and reserve the second and third for the warning and critical levels respectively. By making our check_pg command so generic, we can use it for every action that check_postgres supports. Otherwise, we would have needed a separate command section for each check.

Then, we will add a service check. We will need to add one of these for each check_postgres action that we want to enact. In our example, we only enabled the database_size check and applied the same thresholds that we used when manually invoking the script. By reloading the Nagios configuration files, it will incorporate the new PostgreSQL database size check and apply it to any server that we have in the pg-servers group.

Though the documentation explains all the actions available for check_postgres, it may be inconvenient to refer to it regularly. Though the check_postgres.pl script accepts the usual --help parameter, it has a notable ability as well. If we specify the --man parameter instead, check_postgres will actually display the entire manual. This is similar to investigating the check_postgres man page like this:

man check_postgres

Sometimes, man pages don't get installed properly or are not available for one reason or another. The --man parameter should always work on any system that also contains the perl documentation package.

As check_postgres is developed by Bucardo, their site contains various resources related to its operation. We recommend these links for more information: