- preparation for thread model
- read configuration for munin-node hosts from file
- construction of configuration for each thread done
- filter list of plugins according to supplied regexp
- optimize way how to get plugins from hosts (do not read it every time)
- thread aware signal handler
- handle signal for list of plugin reload
- syslog logging instead basic one
- extended documentation in README.md
- any munin-node could support remote or indirect fetching of plugin's data
- add support for munin-node:remote-node syntax of host parameter
- add process logic to ask for remote node data
- displayname should be used to correctly identify such metrics
* Establishing one carbon connection for all plugins, instead of one connection
for each. This rather drastically improves performance. From a few measurements
this reduces processing time by over 50%
* This can be even further enhanced, to keep one persistent carbon (/munin?)
connection and only re-establish it if it drops (For Future)
* reducing the actual processing time from the interval, so it runs
in more accurate intervals independent of the processing time.
* if processing time takes longer than the interval, we run it instantly
