Created attachment 1165858 [details]
Script that installs and configures collectd in a very basic way for ease of reproducability.

Description of problem:

There is a bug in collectd that causes spammy messages to the system journal when using the exec plugin. Each successful data collection from this plugin causes a log entry. This can mean messages at rates of up to (and even over) once per second depending on the number of instances of the plugin configured and their interval settings.

Please note that although collectd-exec is a plugin, it is included in the main collectd package and just requires configuration to use.

The bug has been acknowledged and already fixed by the collectd project maintainers. See Github issue here: https://github.com/collectd/collectd/issues/664 .


Version-Release number of selected component (if applicable):

(Current Fedora 23 package)
Version: 5.5.1 
Release: 2.fc23


How reproducible:

Easily reproducible. Any correct configuration of the exec plugin for collectd will cause the issue. See the man page here: https://collectd.org/documentation/manpages/collectd-exec.5.shtml


Steps to Reproduce:

1. Configure collectd-exec plugin to generate some metric (see attached for a script that sets up an example configuration).


Actual results:

Observe journal with "journalctl -r -u collectd"
Output:
<snip>
... 0 Success: 1 value has been dispatched.
... 0 Success: 1 value has been dispatched.
... 0 Success: 1 value has been dispatched.
... etc.
</snip>


Expected results:

Only non-success cases should generate a journal message.


Additional info:

The commit that fixes the bug can be seen here:
https://github.com/collectd/collectd/commit/10b10a6718dbb87c6890d4bc821a6e21c8b7b527
It has already been accepted and merged with the project's master.