I think my approach to this would be this: Keep your endpoints simple.
Take a look at what the system comes with. From a quick check, I'm almost certain both CentOS and Fedora ship with rsyslog, which would lead me to believe that the Pi would have that or something very similar. If not, rsyslog does at least have an ARM package.
That native syslog package should almost certainly have the capability to send it's messages (or generally any you might want to specify) to a remote server. I would configure that to push out to a remote server of your choice and do any parsing, filtering, or manipulating server-side. That way you don't have to worry about keeping a package up to date, or ensuring you have a system/application specific configuration. The generic config you'll end up with can, in theory, be applied to anything running linux.
Hope this helps.