- essential, simple and reliable.
- minimum maintenance.
- probes: run probes that gather data from the current or any remote node.
- analyzers: process collected hosts, triggering events (notifications, graphing, etc).
- syncer: send collected and analyzed data to remote locations.
- cronjob: synced dispatch.
- command line: asynced dispatch.
- plugin lookup in shared folders such as
~/.local/share and /usr/local/share.
- analyzer cronjobs detects collected data and dispatch available analyzers.
- outputs: yaml, json or any other format understood by each probe/analyzer pair.
- version field: so analyzers knows by which schema they should process data.
- unique locations per host/probe/analyzer to ease syncing.
- folders: {/etc,/var/lib}/metrics/$HOSTNAME/{probes,analyzers,syncers}/{localhost,$HOST}/$NAME/.
- files: config.yaml (failing back to defaults) and $TIMESTAMP.{json,yaml,rrd,etc}.
metrics <action> <target>
